Wed 6 May 2026

Anthropic comes for the midmarket software spend

Backed by private equity and banking giants, it will build custom AI systems for business bottlenecks There’s gold in midmarket IT spend, and Anthropic - backed by private equity and banking heavyweights and tapping its Claude Partner Network - is coming for it.…

Nvidia must face infringement allegations over AI training in US lawsuit

A US federal judge ruled that Nvidia must face claims regarding the use of copyrighted works to train its AI models, though the court dismissed the vicarious infringement allegation.

Do Venture Capitalists Beat Random Allocation?

arXiv:2605.03980v1 Announce Type: new Abstract: Venture capital outcomes are dominated by a small number of extreme successes, making it difficult to distinguish investor skill from favorable realizations in a highly skewed return distribution. We study this question by comparing empirical VC portfolios to a constrained random benchmark that preserves key portfolio characteristics, including timing, geography, sector composition, and portfolio size, while randomizing individual company selection. Across funding stages, empirical portfolio distributions appear remarkably close to their random benchmarks. We find no evidence that portfolio construction increases the probability of high-multiple outcomes: the right tail remains statistically indistinguishable from random allocation. Deviations in the lower part of the distribution are small and sensitive to the interpretation of zero outcomes, suggesting at most weak evidence of downside improvement. We further introduce a rank-based benchmark distribution to evaluate outperformance at each position in the cross-section. This analysis shows that even the best-performing portfolios do not exceed the outcomes expected for their rank under random sampling. Our results suggest that VC portfolio outcomes are largely consistent with constrained random allocation, highlighting the difficulty of identifying aggregate skill in heavy-tailed investment environments. A similar conclusion holds for the performance of financial analysts in predicting future earnings.

Are AI stocks the new railroad bonds?

To understand what’s going on in the sector, look at 19th-century railroads

Evidence Suggests Generative AI Represents a Structural Boom Rather Than a Market Bubble

Analysis of current investment cycles suggests that scarcity of resources, rather than over-investment, remains the primary constraint on AI development. The narrative of a bubble is increasingly contradicted by sustained demand and infrastructure bottlenecks.

OpenAI exec says company hopes to burn $50B of somebody else's money on compute this year

If the numbers are large enough, perhaps we won't question the math An executive for ChatGPT maker OpenAI said in court testimony on Tuesday that the AI model developer expects to burn $50 billion on computing power before the end of the year.…

JPMorgan and BlackRock bosses play down talk of AI bubble

Dimon and Fink upbeat in separate comments about demand for the technology as Wall Street funds sector’s spending

Oaktree BDC Marks Down Software Loans, Flags 26% AI Exposure

Oaktree Capital Management cut the value of one of its private credit funds by almost 4% as the firm marked down its software assets.

SAP To Acquire Prior Labs And Invest Over €1 Billion To Build Frontier AI Lab In Europe

SAP SE has entered into a definitive agreement to acquire Prior Labs, the pioneer of Tabular Foundation Models, in a deal that commits the enterprise software giant to investing more than €1 billion over the next four years to scale Prior Labs into a globally leading frontier AI research lab.

“The math is not mathing”: How AI bubble fears are changing Canadian VC’s investment approach | BetaKit

Leaders from Wittington, McRock, and IRV clash on whether AI is a bubble, but agree some things do not add up.

AI News Digest, May 5: Private Equity Becomes the AI Deployment Channel

Private equity AI deployment got its own $11.5B vehicle this week. What HR and ops leaders do now, plus sovereign-cloud AI and India state policy.

SAP bets $1.16B on 18-month-old German AI lab and says yes to NemoClaw

SAP is investing $1.16 billion into a young German AI lab, signaling a major commitment to the NemoClaw project.

Can Britain’s star tech investor dodge the SaaSpocalypse?

Delayed IPO of €19bn software group Visma was meant to be a crowning moment for PE group Hg — and London

South Korean chipmakers emerge in ranking of companies with biggest global AI revenue growth | News | Eco-Business | Asia Pacific

SK Hynix and Samsung Electronics are among the world's fastest-growing AI revenue companies on the back of surging demand for high-bandwidth memory and AI chips, according to investing research platform BestBrokers.

Memory Chips Ignite Global Tech Stock Frenzy; China's STAR 50 Index Surges Over 9% — BigGo Finance

China's A-share market kicked off with a bang on the first trading day after the May Day holiday, as the memory chip and GPU sectors triggered a wave of circuit

Supermicro CEO insists ‘no one’ beyond indicted employees were involved in alleged $2.5 billion smuggling scheme

CEO and Chairman Charles Liang said Supermicro’s relationship with vendors including Nvidia, AMD, Intel, and Broadcom was “strong” despite the accusations against a Supermicro cofounder. Supermicro stock rose 17% in after-hours trading.

Opinion | Competition for Pentagon AI contracts ensures troops get best tools - The Washington Post

Silicon Valley has been too ambivalent about American power for the first quarter of this century. But deals announced Friday by the Pentagon show how that’s changing for the better, aided by intense competition within the software industry.

AI Labs: Are Anthropic really the good guys?

Dario Amodei casts his company as the good guys of the AI race. Will that last?

Tech Giants Just Made Every Business Their Business | PYMNTS.com

Announcements from Amazon, OpenAI and Anthropic point to a model where enterprise technology is deployed across entire portfolios at once.

Apple's chip talks with Intel and Samsung signal a structural shift in how Big Tech thinks about supply chain risk | NASDAQ:AAPL, XETRA:APC

The iPhone maker's exploratory discussions are less about finding a better manufacturer than about ensuring it never again loses iPhone revenue to a single...

Make way for agents

AI developers are increasingly building software specifically for agents rather than humans. This shift could change tech competition from interface design to controlling the APIs and data agents need to function.

Consumer Choice Over Shopping Baskets

arXiv:2511.11846v2 Announce Type: replace Abstract: I introduce a novel approach to structural modelling and estimation of continuous demand systems, utilising consideration sets to analyse differentiated products markets with very large choice sets and purchases over multiple goods, multiple units, and across product categories. I apply it to study intra-store competition in the Portuguese supermarket industry between 2020 and 2023, during which the country faced the COVID pandemic. Anonymised transaction-level point-of-sale data is sufficient to estimate price elasticities across almost 30 000 goods and more than 500 product categories. Results suggest mark-ups remained stable throughout the sample period, with a short-lived, slight increase post-pandemic observed only in the highest-mark-up-percentile goods. The implied mark-ups match observed price volatility, profit margin surveys, as well as reports on shifting consumer tastes during the sample period.

The Distillation Panic

This article explains why model distillation is an important AI training technique and warns against conflating legitimate distillation with API abuse or competitive extraction.

Microsoft gives up on Xbox Copilot AI

Microsoft is winding down its Xbox Copilot AI project as the company shifts its gaming strategy toward optimization.

Who Wins in the AI Services Economy? Join Our Webinar Discussion - Newsweek

“The opportunity is increasing,” Noshir Kaka says. “But so is the competition.” AI Impact Forum looks at what’s next.

Apple could let you pick a favorite AI model in iOS 27

Reports suggest future iOS updates may allow users to select their preferred third-party AI models for system integration.

China's cloud providers raise AI prices as model usage surges

Rising demand for generative AI is prompting major Chinese cloud providers to raise prices for large-scale services and data products, with implications for global AI users and developers, as usage-based billing and efficiency become central to costs and deployment decisions across international ...

What you'll pay for AI agents will be wildly variable and unpredictable | ZDNET

A test of leading AI agents found vastly different amounts of tokens consumed with no transparency and no guarantees of success.

Coalition calls for ban on surveillance pricing, electronic shelf labels in New York

Krutrim Is Pivoting From Frontier AI Models to Cloud Infrastructure and the Move Reveals How National AI Champions Actually Monetize in Emerging Markets – Startup Fortune

Krutrim, India's first generative AI unicorn founded by Bhavish Aggarwal at a $1 billion valuation after a $50 million Matrix-led round, is shifting

AI startup Sierra raises $950 million at $15 billion valuation - The Economic Times

AI startup Sierra has secured a significant funding round of $950 million. This investment values the company at over $15 billion. Sierra plans to use the funds to expand its platform. The company aims to become a global standard for AI-driven customer experience.

Paris-based Lithosquare raises €21.4 million to accelerate transition-critical mineral discovery with Geology AI

Lithosquare, a Paris-based startup that deploys Geology AI and geologist-led intelligence to amplify and accelerate the discovery of transition-critical minerals, has raised €21.3 million ($25 million) in fresh funding.  The round was co-led by World Fund and Kindred Capital, with participation from Daphni, Omnes Capital and Ovni Capital. The company plans to use this funding […]

Human-Provenance Verification should be Treated as Labor Infrastructure in AI-Saturated Markets

arXiv:2605.03210v1 Announce Type: cross Abstract: We argue that AI-saturated markets are likely to create Veblen-good premiums, which we term human-provenance premiums, for verified human presence, and hence AI governance should treat human-provenance verification as labor infrastructure. Generative and agentic AI systems lower the cost of many standardized cognitive, creative, and coordination tasks, weakening the scarcity premiums that have supported much middle-tier knowledge work. We argue that this pressure may produce an asymmetric barbell-shaped structure of value capture in advanced economies: high-volume synthetic production controlled by owners of AI infrastructure at one pole, and scarce, high-status human labor valued for verified human presence at the other. We advance three claims. First, AI compresses the value of standardized middle-tier labor by making good-enough synthetic substitutes scalable at low marginal cost, hollowing out the middle of the skill distribution currently categorized by knowledge work. Second, this compression reallocates demand for human labor toward work valued for its visible human character. We term this performative humanity and distinguish three forms of labor: relational presence, aesthetic provenance, and accountability. Third, as these premiums depend on credible verification, AI governance should treat human-provenance systems as labor infrastructure rather than as luxury authenticity labels. To evaluate hybrid human-AI work, we propose constitutive human presence as the relevant standard: human labor retains premium value when human judgment, attention, accountability, authorship, or relational participation is not incidental to the output but constitutive of what is being purchased.

Cheap Expertise: Mapping and Challenging Industry Perspectives in the Expert Data Gig Economy

arXiv:2605.03295v1 Announce Type: new Abstract: Demand for expert-annotated data on the part of leading AI labs has created an expert gig economy with the potential to reshape white collar work and society's understanding of expertise. In this research, we study the vision for the future of expertise described in the public communication of five industry data annotation organizations and their CEOs, as reflected on social media feeds and public appearances on podcasts. We find that the industry envisions AI expertise as cheap, meaning that it can offer a better return on investment than human expertise. Human expertise, meanwhile, is viewed as an extractable resource, the value of which can be judged relative to AI expertise. Finally, institutional expertise (such as that created or possessed by universities and corporations) is viewed as in need of liberation or reform, such that it can be incorporated into the latest artificial intelligence systems. Our findings have implications for human experts, whose professional lives may be transformed and revalued by this industry, as well as for societal institutions that mediate expertise. We close this work with a series of provocations intended to elicit consideration of how society can best approach an AI-driven expert gig economy and the cheap expertise it intends to produce.

Inteligencia artificial y empleo en Espa\~na: una aproximaci\'on territorial y de g\'enero a la exposici\'on laboral

arXiv:2512.23059v3 Announce Type: replace Abstract: The diffusion of artificial intelligence, particularly generative models, is expected to transform labor markets in uneven ways across sectors, territories, and social groups. This paper proposes a methodological framework to estimate the potential exposure of employment to AI using sector based data, addressing the limitations of occupation centered approaches in the Spanish context. By constructing an AI CNAE incidence matrix and applying it to provincial employment data for the period 2021 to 2023, we provide a territorial and gender disaggregated assessment of AI exposure across Spain. The results reveal stable structural patterns, with higher exposure in metropolitan and service oriented regions and a consistent gender gap, as female employment exhibits higher exposure in all territories. Rather than predicting job displacement, the framework offers a structural perspective on where AI is most likely to reshape work and skill demands, supporting evidence based policy and strategic planning.

Public and private markets vie for gains from AI job disruption

Corporate leaders are betting that automation will produce outsized returns

Coinbase to cut jobs and rebuild the group as an ‘intelligence’

Crypto exchange’s chief says AI is speeding up its processes, meaning fewer employees are needed

The Federal Safety Net Isn’t Ready for Artificial Intelligence

As fears of A.I.-driven job losses mount, economists warn that unemployment benefits and other programs to help displaced workers aren’t sufficient.

11 companies, including IBM and Coinbase, that have said they're doing AI-related layoffs

A number of companies, including Snap and Coinbase, have attributed recent reductions in staff to AI.

Workers are training AI and robots to replace them. It could end badly | The Independent

Humans have been working themselves out of a job for millennia, writes Anthony Cuthbertson. But new tactics from tech firms have experts warning that we are ‘sleepwalking' into a labour market overrun by AI and humanoid robots

Google DeepMind Workers Vote to Unionize Over Military AI Deals

Employees at Google DeepMind have voted to unionize, citing concerns over the company's involvement in military AI contracts.

Brainrot: Deskilling and Addiction are Overlooked AI Risks

arXiv:2605.03512v1 Announce Type: new Abstract: The scope of AI safety and alignment work in generative artificial intelligence (GenAI) has so far mostly been limited to harms related to: (a) discrimination and hate speech, (b) harmful/inappropriate (violent, sexual, illegal) content, (c) information hazards, and (d) use cases related to malicious actors, such as cybersecurity, child abuse, and chemical, biological, radiological, and nuclear threats. The public conversation around AI, on the other hand, has also been focusing on threats to our cognition, mental health, and welfare at large, related to over-relying on new technologies, most recently, those related to GenAI. Examples include deskilling associated with cognitive offloading and the atrophy of critical thinking as a result of over-reliance on GenAI systems, and addiction associated with attachment and dependence on GenAI systems. Such risks are rarely addressed, if at all, in the AI safety and alignment literature. In this paper, we highlight and quantify this discrepancy and discuss some initial thoughts on how safety and alignment work could address cognitive and mental health concerns. Finally, we discuss how information campaigns and regulation can be used to mitigate such prominent risks.

Beyond Distributive Justice: Hermeneutical Fairness in Ad Delivery

arXiv:2605.03419v1 Announce Type: new Abstract: Fairness in online advertising is often formalized as a distributive justice problem, aiming to ensure that impressions, opportunities, or outcomes are allocated comparably across protected groups. Yet online advertising can still produce harms arising from ads' content and from how recipients interpret and uptake them. To capture this dimension, we draw on Miranda Fricker's notion of hermeneutical injustice. We model ad delivery as a mechanism that distributes interpretative resources and can fail in two ways: relevant concepts can be withheld through systematic under-exposure, leading to hermeneutical deprivation; and recipients may experience hermeneutical distortions when saturated with low-uptake or skewed framings. Grounded in exploratory correlational patterns from the AIDS Advertising Evaluation surveys (1986-1987), we introduce a group-level hermeneutical fairness constraint and a hermeneutically aware utility cost. We integrate them into a benchmark, utility-driven ad allocation framework that already enforces distributive justice, yielding a distributively fair, hermeneutically aware framework that prevents deprivation and distortion from concentrating within protected groups. Through controlled simulations, we explore trade-offs between economic utility, classical distributive fairness constraints, and hermeneutical cost. The results show that purely utility-based allocation drives under-delivery to the disadvantaged group. When the hermeneutical stakes of withholding ads are high, distributive constraints reduce hermeneutical cost at modest utility loss. Conversely, weighting hermeneutical cost without distributive constraints can yield policies concentrated on the disadvantaged group. These findings motivate expanding fairness analyses of online advertising beyond distributive notions to include epistemic conditions of interpretation and uptake.

Character.AI sued by Pa. over alleged doctor impersonation by chatbot

Pennsylvania's Department of State has sued chatbot developer Character.AI, alleging the company misrepresented its companion chatbots as licensed medical professionals.

Understanding Emergent Misalignment via Feature Superposition Geometry

arXiv:2605.00842v1 Announce Type: new Abstract: Emergent misalignment, where fine-tuning on narrow, non-harmful tasks induces harmful behaviors, poses a key challenge for AI safety in LLMs. Despite growing empirical evidence, its underlying mechanism remains unclear. To uncover the reason behind this phenomenon, we propose a geometric account based on the geometry of feature superposition. Because features are encoded in overlapping representations, fine-tuning that amplifies a target feature also unintentionally strengthens nearby harmful features in accordance with their similarity. We give a simple gradient-level derivation of this effect and empirically test it in multiple LLMs (Gemma-2 2B/9B/27B, LLaMA-3.1 8B, GPT-OSS 20B). Using sparse autoencoders (SAEs), we identify features tied to misalignment-inducing data and to harmful behaviors, and show that they are geometrically closer to each other than features derived from non-inducing data. This trend generalizes across domains (e.g., health, career, legal advice). Finally, we show that a geometry-aware approach, filtering training samples closest to toxic features, reduces misalignment by 34.5%, substantially outperforming random removal and achieving comparable or slightly lower misalignment than LLM-as-a-judge-based filtering. Our study links emergent misalignment to feature superposition, providing a basis for understanding and mitigating this phenomenon.

Organizational Theory Is Essential for Managing the Complexity of Multi-Agent AI Systems

Current approaches to agentic systems rely too heavily on technical control planes, ignoring the necessary organizational frameworks. Effective deployment requires integrating management principles like decision rights and spans of control.

AWS lets agents drive its virtual cloudy desktops - which could cost 500,00 tokens per click

Vendor benchmark finds APIs let you do the job faster and cheaper Amazon Web Services has let AI agents loose in its cloudy WorkSpaces virtual PCs.…

Anthropic wants Claude to play with money, unleashes finance agents

Always bet on backpropagation If you've ever read Anthropic's disclaimer that responses generated by Claude may contain mistakes and thought, "That's what I need to spice up financial operations," you're in luck.…

Virtual Speech Therapist: A Clinician-in-the-Loop AI Speech Therapy Agent for Personalized and Supervised Therapy

arXiv:2605.01101v1 Announce Type: new Abstract: This paper develops Virtual Speech Therapist (VST), an intelligent agent-based platform that streamlines stuttering assessment and delivers customized therapy planning through automated and adaptive AI-driven workflows. VST integrates state-of-the-art deep learning-based stuttering classification, and multi-agent large language model (LLM) reasoning to support evidence-based clinical decision-making. The VST begins with the acquisition and feature extraction of patient speech samples, followed by robust classification of stuttering types. Building on these outputs, VST initiates an agentic reasoning process in which specialized LLM agents autonomously generate, critique, and iteratively refine individualized therapy plans. A dedicated critic agent evaluates all generated therapy plans to ensure clinical safety, methodological soundness, and alignment with peer-reviewed evidence and established professional guidelines. The resulting output is a comprehensive, patient-specific therapy draft intended for clinician review. Incorporating clinician feedback, the system then produces a finalized therapy plan suitable for patient delivery, thereby maintaining a clinician-in-the-loop paradigm. Experimental evaluation by expert speech therapists confirms that VST consistently generates high-quality, evidence-based therapy recommendations. These findings demonstrate the system's potential to augment clinical workflows, reduce clinician burden, and improve therapeutic outcomes for individuals with speech impairments. An interactive user interface for the proposed system is available online at: https://vocametrix.com/ai/stuttering-therapy-planning-agent , facilitating real-time stuttering assessment and personalized therapy planning.

IBM asks DBAs to trust AI to act on their behalf

With help from Google and Intel, Big Blue brings new automation to Db2 IBM has added support for Google Vertex AI and Intel Gaudi to boost the AI-based management of its stalwart Db2 database.…

AI agent governance becomes boardroom risk for contact centers - Outsource Accelerator

Enterprise contact centers are deploying AI agents from Microsoft, SAP,Salesforce and other vendors at a pace that has outstripped basic governance.

Uber Shares What Happens When 1,500 AI Agents Hit Production

Uber details the operational challenges and outcomes of deploying 1,500 autonomous AI agents into their production environment.

JumpCloud releases ‘Agentic IAM Pulse Report’ - MacTech.com

JumpCloud Inc., an U.S.-based enterprise software company, has released its Agentic IAM Pulse Report, which examines how organizations are rapidly expanding their use of AI agents across internal and business-critical workflows and the growing disconnect between deployment scale, identity management

Meta plans advanced ‘agentic’ AI assistant for consumers

Social media platform invests in equivalent to OpenClaw that aims to seamlessly carry out everyday tasks for users

Brit mathematician lets AI agent loose with credit card – cue password leaks, CAPTCHA chaos and more

Professor Fry's AI experiment shows light and dark sides of agentic tech.

Towards Multi-Agent Autonomous Reasoning in Hydrodynamics

arXiv:2605.01102v1 Announce Type: new Abstract: Single-agent systems (SAS) have become the default pattern for LLM-driven scientific workflows, but routing planning, tool use, and synthesis through a single context window comes with a well-known cost: as tool specifications and observational traces accumulate, the effective context available for each decision shrinks, and end-to-end reliability suffers. We present a multi-agent system (MAS) prototype for hydrodynamics in which specialized agents are coordinated through a Layer Execution Graph (LEG). A planner agent constructs query-specific execution topologies from natural-language routing heuristics that capture domain knowledge without hard-coding it as rigid control logic; specialist agents operate under strict tool allowlists and occupy complementary data-class roles. Between layers, consolidator agents fuse parallel outputs into concise briefs, and a reporter agent synthesizes the final response, while the runtime logs provenance for every tool invocation to support auditability. All benchmarks, ablations, and stress tests use Claude Sonnet~4.6 as the backbone model for both specialist and general-purpose agents. Evaluated on 37 queries spanning six complexity categories, the prototype achieves 93.6% factual precision with a 100% pass rate. Accuracy remains above 90% across runs from single-threaded to five independent parallel tracks, and under simulated loss of individual data sources the system degrades gracefully, still returning substantive partial answers. Together, these results suggest that planner-guided, graph-structured multi-agent orchestration can meaningfully alleviate the context-saturation bottlenecks that constrain monolithic single-agent architectures.

Anthropic in discussions to by inference chips from UK startup Fractile – report

Anthropic has held discussions with Fractile to buy inference chips from the UK-based startup when its hardware becomes available next year, per a report from The Information. Citing two people with knowledge of the situation, The Information noted that talks between the two companies were in the early stages, and details about the size of […]

Astera speaks softly and carries a big switch

High-speed connectivity without NVLink baggage Astera Labs unveiled an alternative to Nvidia's NVSwitch for building rack-scale AI systems on Tuesday, claiming it will work with nearly any accelerator.…

AMD earnings call: AI is turning CPUs back into the main event

AMD's fiscal first-quarter 2026 earnings call was not just a victory lap for another data center beat. It was a strategic argument from management: AI infrastructure is no longer only an accelerator story. It is becoming a full compute-platform story, where CPUs, GPUs, memory, software, and ...

GlobalFoundries turns three-continent footprint into geopolitical hedge

In a recent assessment of the semiconductor landscape, GlobalFoundries emphasized the critical need for supply chain resilience amid a "fragmented geopolitical environment." The company is positioning its three-continent manufacturing presence—spanning the US, Germany, and Singapore—as ...

Blue Owl Data Center Operator Stack Is Said to Consider $30 Billion Sale of Asia Operations

Stack Infrastructure Inc., a data center company owned by Blue Owl Capital, is considering options including a sale of its Asia operations, according to people familiar with the matter.

Microsoft to stop taking reservations for 17 Azure VM flavours, kill 13 in 2028

Haswell’s had its day and Skylake and Cascade Lake are draining away.

AI Datacenter Liquid Cooling Market to Reach USD 17.8 Billion by 2036 as Hyperscale AI Infrastructure Drives Thermal Management Transformation | Morningstar

This growth reflects a structural transformation in datacenter thermal management, where operators are moving beyond traditional air-cooling architectures toward liquid-based cooling systems capable of supporting ultra-high-density AI racks and GPU training clusters. As AI accelerators continue to increase rack-level power densities, liquid cooling is becoming a critical infrastructure requirement for maintaining thermal efficiency, reducing energy consumption...

Inside Amazon Web Services' plan to make networking disappear

A look inside the AWS networking lab in Cupertino reveals the company's efforts to simplify cloud networking.

AI data centers head for the ocean

The industry is exploring moving AI data centers to the ocean to address cooling and energy demands.

Miami startup Subquadratic claims 1,000x AI efficiency gain with SubQ model; researchers demand independent proof.

A little-known Miami-based startup called Subquadratic emerged from stealth on Tuesday with a sweeping claim: that it has built the first large language model to fully escape the mathematical constraint that has defined — and limited — every major AI system since 2017. The company claims its first model, SubQ 1M-Preview, is the first LLM built on a fully subquadratic architecture — one where compute grows linearly with context length. If that claim holds, it would be a genuine inflection point in how AI systems scale. At 12 million tokens, the company says, its architecture reduces attention compute by almost 1,000 times compared to other frontier models — a figure that, if validated independently, would dwarf the efficiency gains of any existing approach. The company is also launching three products into private beta: an API exposing the full context window, a command-line coding agent called SubQ Code, and a search tool called SubQ Search. It has raised $29 million in seed funding from investors including Tinder co-founder Justin Mateen, former SoftBank Vision Fund partner Javier Villamizar, and early investors in Anthropic, OpenAI, Stripe, and Brex. The New Stack reported that the raise values the company at $500 million. The numbers Subquadratic is publishing are extraordinary. The reaction from the AI research community has been, to put it mildly, mixed — ranging from genuine curiosity to open accusations of vaporware. Understanding why requires understanding what the company claims to have solved, and why so many prior attempts to solve the same problem have fallen short. The quadratic scaling problem has shaped the economics of the entire AI industry Every transformer-based AI model — which includes virtually every frontier system from OpenAI, Anthropic, Google, and others — relies on an operation called "attention." Every token is compared against every other token, so as inputs grow, the number of interactions — and the compute required to process them — scales quadratically. In plain terms: double the input size, and the cost doesn't double. It quadruples. This relationship has shaped what gets built and what doesn't. The industry standard is 128,000 tokens for many AI models and up to 1 million tokens for frontier cloud models such as Claude Sonnet 4.7 and Gemini 3.1 Pro.  Even at those sizes, the cost of processing long inputs becomes punishing. The industry built an elaborate stack of workarounds to cope. RAG systems use a search engine to pull a small number of relevant results before sending them to the model, because sending the full corpus isn't feasible. Developers layer retrieval pipelines, chunking strategies, prompt engineering techniques, and multi-agent orchestration systems on top of models — all to route around the fundamental constraint that the model itself can't efficiently process everything at once. Subquadratic's argument is that these workarounds are expensive, brittle, and ultimately limiting. As CTO Alexander Whedon told SiliconANGLE in an interview, "I used to manually curate prompts and retrieval systems and evals and conditional logic to chain together the workflows. And I think that that is kind of a waste of human intelligence and also limiting to the product quality." Subquadratic's fix is deceptively simple: stop doing the math that doesn't matter The company's approach, called Subquadratic Sparse Attention or SSA, is built on a straightforward premise: most of the token-to-token comparisons in standard attention are wasted compute. Instead of comparing every token to every other token, SSA learns to identify which comparisons actually matter and computes attention only over those positions. Crucially, the selection is content-dependent — the model decides where to look based on meaning, not on fixed positional patterns. This allows it to retrieve specific information from arbitrary positions across a very long context without paying the quadratic tax. The practical payoff scales with context length — exactly the inverse of the problem it's trying to solve. According to the company's technical blog, SSA achieves a 7.2x prefill speedup over dense attention at 128,000 tokens, rising to 52.2x at 1 million tokens. As Whedon put it: "If you double the input size with quadratic scaling laws, you need four times the compute; with linear scaling laws, you need just twice." The company says it trained the model in three stages — pretraining, supervised fine-tuning, and a reinforcement learning stage specifically targeting long-context retrieval failures — teaching the model to aggressively use distant context rather than defaulting to nearby information, a subtle failure mode that quietly degrades performance in existing systems. Three benchmarks paint a strong picture, but what they leave out may matter more On the surface, SubQ's benchmark numbers are competitive with or superior to models built by organizations spending billions of dollars. On SWE-Bench Verified, it scored 81.8% compared to Opus 4.6's 80.8% and DeepSeek 4.0 Pro's 80.0%. On RULER at 128,000 tokens, a standard benchmark for reasoning over extended inputs, SubQ scored 95% — edging out Claude Opus 4.6 at 94.8%. On MRCR v2, a demanding test of multi-hop retrieval across long contexts, SubQ posted a third-party verified score of 65.9%, compared with Claude Opus 4.7 at 32.2%, GPT-5.5 at 74%, and Gemini 3.1 Pro at 26.3%. But several details warrant scrutiny. The benchmark selection is narrow — exactly three tests, all emphasizing long-context retrieval and coding, the precise tasks SubQ is designed for. Broader evaluations across general reasoning, math, multilingual performance, and safety have not been published. The company says a comprehensive model card is "coming soon." According to The New Stack, each benchmark model was run only once due to high inference cost, and the SWE-Bench margin is, as the company's own paper acknowledges, "harness as much as model." In benchmark methodology, single runs without confidence intervals leave room for variance. There is also a significant gap between SubQ's research results and its production model. On MRCR v2, the company reported a research score of 83 — but the third-party verified production model scored 65.9. That 17-point gap between the lab result and the shipping product is notable and largely unexplained. Subquadratic also told SiliconANGLE that on the RULER 128K benchmark, SubQ scored 95% accuracy at a cost of $8, compared with 94% accuracy and about $2,600 for Claude Opus — a remarkable cost claim. But the company has not publicly disclosed specific API pricing, making it impossible to independently verify the cost-per-task comparisons. The AI research community's verdict ranges from 'genuine breakthrough' to 'AI Theranos' Within hours of the announcement, the AI research community erupted into a debate that crystallized around a single question: Is this real? AI commentator Dan McAteer captured the binary mood in a widely shared post: "SubQ is either the biggest breakthrough since the Transformer... or it's AI Theranos." The comparison to the infamous blood-testing fraud company may be unfair, but it reflects the scale of the claims being made. Skeptics zeroed in on several pressure points. Prominent AI engineer Will Depue initially noted that SubQ is "almost surely a sparse attention finetune of Kimi or DeepSeek," referring to existing open-source models. Whedon confirmed this on X, writing that the company is "using weights from open-source models as a starting point, as a function of our funding and maturity as a company." Depue later escalated his criticism, writing that the company's O(n) scaling claims and the speedup numbers "don't seem to line up" and called the communication "either incredibly poorly communicated or just not real." Others raised structural questions. One developer noted that if SubQ truly reduces compute by 1,000x and costs less than 5% of Opus, the company should have no trouble serving it at scale — so why gate access through an early-access program? Developer Stepan Goncharov called the benchmarks "very interesting cherry-picked benchmarks," while another commenter described them as "suspiciously perfect." But not everyone was dismissive. AI researcher John Rysana pushed back on the Theranos framing, writing that the work is "just subquadratic attention done well which is very meaningful for long context workloads," and that "odds of it being BS are extremely low." Linus Ekenstam, a tech commentator, said he was "extremely intrigued to see the real-world implications" particularly for complex AI-powered software. Magic.dev made strikingly similar claims two years ago — and then went quiet Perhaps the most pointed critique of SubQ's launch comes not from its specific claims but from recent history. Magic.dev announced a 100-million-token context-window model in August 2024, with a claimed 1,000x efficiency advantage, and raised roughly $500 million on the strength of those claims. As of early 2026, there is no public evidence of LTM-2-mini being used outside Magic. The parallels are uncomfortable. Both companies claimed massive context windows. Both touted roughly 1,000x efficiency gains. Both targeted software engineering as their primary use case. And both launched with limited external access. The broader research landscape reinforces the caution. Kimi Linear, DeepSeek Sparse Attention, Mamba, and RWKV all promised subquadratic scaling, and all faced the same problem: architectures that achieve linear complexity in theory often underperform quadratic attention on downstream benchmarks at frontier scale, or they end up hybrid — mixing subquadratic layers with standard attention and losing the pure scaling benefits. A widely cited LessWrong analysis argued that these approaches "are all better thought of as 'incremental improvement number 93595 to the transformer architecture'" because practical implementations remain quadratic and "only improve attention by a constant factor." Subquadratic is directly aware of this history. Its own technical blog specifically addresses each prior approach — fixed-pattern sparse attention, state space models, hybrid architectures, and DeepSeek Sparse Attention — and argues that SSA avoids their tradeoffs. Whether it actually does remains an empirical question that only independent evaluation can settle. A five-time founder, a former Meta engineer, and $29 million to prove the doubters wrong The team behind the claims matters in evaluating them. CEO Justin Dangel is a five-time founder and CEO with a track record across health tech, insurancetech, and consumer goods, and his companies have scaled to hundreds of employees, attracted institutional backing, and reached liquidity. CTO Alexander Whedon previously worked as a software engineer at Meta and served as Head of Generative AI at TribeAI, where he led over 40 enterprise AI implementations. The team includes 11 PhD researchers with backgrounds from Meta, Google, Oxford, Cambridge, ByteDance, and Adobe. That is a credible collection of talent for an architecture-level research effort. But neither co-founder has published foundational AI research, and the company has not yet released a peer-reviewed paper. The technical report is listed as "coming soon." The funding profile is unusual for a company making frontier AI claims. Subquadratic raised $29 million at a reported $500 million valuation — a steep price for a seed-stage company with no publicly available model, no peer-reviewed research, and no disclosed revenue. The investor base, led by Tinder co-founder Mateen and former SoftBank partner Villamizar, skews toward consumer tech and growth investing rather than deep technical AI research. The company is not open-sourcing its weights but plans to offer training tools for enterprises to do their own post-training, and has set a 50-million-token context window target for Q4. The real test for SubQ isn't benchmarks — it's whether the math survives independent scrutiny Strip away the marketing language and the social media drama, and the underlying question Subquadratic is asking is genuinely important: Can AI systems break free of quadratic scaling without sacrificing the quality that makes them useful? The stakes are enormous. If attention can be made truly linear without degrading retrieval and reasoning, the economics of AI shift fundamentally. Enterprise applications that today require elaborate retrieval pipelines — processing entire codebases, contracts, regulatory filings, medical records — become single-pass operations. The billions of dollars currently spent on RAG infrastructure, context management, and agentic orchestration become partially redundant.  Whedon's willingness to engage publicly with technical criticism — posting a technical blog within hours of pushback — suggests a team that understands it needs to show its work, not just describe it. And to its credit, the company acknowledged openly that it builds on open-source foundations and that its model is smaller than those at the major labs. Every frontier model in 2026 advertises a context window of at least a million tokens, but almost none of them are actually great at making use of all that information. The gap between a nominal context window and a functional one — between what a model accepts and what it reliably reasons over — remains one of the most important unsolved problems in AI. Subquadratic says it has closed that gap. If independent evaluation confirms that claim, the implications would ripple far beyond a single startup's valuation. If it doesn't, the company joins a growing list of long-context promises that sounded revolutionary on launch day and unremarkable six months later. In computing, every fundamental constraint eventually falls. When it does, the breakthrough never comes from the direction the industry expected. The question hanging over Subquadratic is whether a team of 11 PhDs and a $29 million seed round actually found the answer that has eluded organizations spending thousands of times more — or whether they just found a better way to describe the problem.

Performance Benchmarks Indicate Rapid Advancement in Commodity AI Model Capabilities

Recent performance data shows that free-tier models are now reaching capability levels that were previously exclusive to top-tier paid models. This rapid commoditization of high-end intelligence has significant implications for market competition.

Conned by a chatbot

Like tricksters, LLMs have perfected the art of plausibility

The Universal Utility of Large Language Models Challenges Traditional Tech Adoption Patterns

LLMs exhibit a unique, broad-spectrum effectiveness across disparate domains, from finance to specialized technical tasks. This versatility distinguishes them from traditional software, complicating standard ROI and deployment models.

Games people — and machines — play: Untangling strategic reasoning to advance AI

Researchers are working to untangle strategic reasoning in AI to improve how machines handle complex, multi-step decision-making tasks.

Bun posts Rust porting guide, says rewrite is still half-baked

Zig's no-AI policy is at odds with view that most open source code will be AI-written in future.

New Bounds for Zarankiewicz Numbers via Reinforced LLM Evolutionary Search

arXiv:2605.01120v1 Announce Type: new Abstract: The Zarankiewicz number $\textbf{Z}(m, n, s, t)$ is the maximum number of edges in a bipartite graph $G_{m, n}$ such that there is no complete $K_{s, t}$ bipartite subgraph. We determine for the first time the exact values of three Zarankiewicz numbers: $\textbf{Z}(11, 21, 3, 3)=116$, $\textbf{Z}(11, 22, 3, 3)=121$, and $\textbf{Z}(12, 22, 3, 3)=132$. We further establish lower bounds for 41 more Zarankiewicz numbers, including several that are within one edge of the best known upper bound, and we match the established value in four more closed cases. Our results are obtained using OpenEvolve, an open-source evolutionary algorithm based on Large Language Models (LLMs) that iteratively improves algorithms for generating mathematical constructions by optimizing a reward signal which we tailored for this specific problem. These findings provide new extremal graph constructions and demonstrate the potential of LLM-guided evolutionary search to contribute to mathematical research. In addition to presenting the resulting constructions, we report the generation algorithms produced, describe the relevant implementation details, and provide our computational costs. Our costs are remarkably low, at less than \$30 for each Zarankiewicz parameter combination, showing that LLM-guided evolutionary search can be an inexpensive, reproducible, and accessible tool for discovering new combinatorial constructions.

SEBI Steps Up AI Cybersecurity Measures with New Task Force | Headlines

The Securities and Exchange Board of India (SEBI) has issued an advisory to warn against the evolving cybersecurity threats posed by advanced artificial intelligence (AI) tools such as Anthropic's Mythos. To address these risks, SEBI has established a special task force named cyber-suraksha.ai. This task force, comprising representatives from various market infrastructure ...

A Low-Latency Fraud Detection Layer for Detecting Adversarial Interaction Patterns in LLM-Powered Agents

arXiv:2605.01143v1 Announce Type: new Abstract: Large Language Model (LLM)-powered agents demonstrate strong capabilities in autonomous task execution, tool use, and multi-step reasoning. However, their increasing autonomy also introduces a new attack surface: adversarial interactions can manipulate agent behavior through direct prompt injection, indirect content attacks, and multi-turn escalation strategies. Existing defense strategies focus on prompt-level filtering and rule-based guardrails, which are often insufficient when risk emerges gradually across interaction sequences. In this work, we propose a complementary defense mechanism: a low-latency fraud detection layer for detecting adversarial interaction patterns in LLM-powered agents. Instead of determining whether a single prompt is malicious, our approach models risk over interaction trajectories using structured runtime features derived from prompt characteristics, session dynamics, tool usage, execution context, and fraud-inspired signals. The detection layer can be implemented using lightweight models leading to low-latency real-time deployments. To evaluate the framework, we construct a synthetic corpus of 12,000 multi-turn agent interactions generated from parameterized templates that simulate realistic agentic workflows. Using 42 structured features and an XGBoost classifier, our detector achieves over 9 times faster than LLM-based detectors. Through the experiment and ablation studies, our work suggests that interaction-level behavioral detection should become a core component of deployment-time defense for LLM-powered agents.

One command turns any open-source repo into an AI agent backdoor

A new vulnerability called OpenClaw allows a single command to turn open-source repositories into AI agent backdoors, highlighting a gap in supply-chain security.

Opinion | AI-powered cyberattack threats are growing. Here's how to combat them. - The Washington Post

Artificial intelligence is tearing down cyberdefenses. Here’s what the government can do to protect Americans.

NHS to close-source hundreds of GitHub repos over AI, security concerns

Healthcare giant's maintainers handed May deadline to enact the change.

Effect-Transparent Governance for AI Workflow Architectures: Semantic Preservation, Expressive Minimality, and Decidability Boundaries

arXiv:2605.01030v2 Announce Type: new Abstract: We present a machine-checked formalization of structurally governed AI workflow architectures and prove that effect-level governance can be imposed without reducing internal computational expressivity. Using Interaction Trees in Rocq 8.19, we define a governance operator G that mediates all effectful directives, including memory access, external calls, and oracle (LLM) queries. Our development compiles with 0 admitted lemmas and consists of 36 modules, ~12,000 lines of Rocq, and 454 theorems. We establishseven properties: (P1) governed Turing completeness, (P2) governed oracle expressivity, (P3) a decidability boundary in which governance predicates are total and closed under Boolean composition while semantic program properties remain non-trivial and undecidable by governance, (P4) goal preservation for permitted executions, (P5) expressive minimality of primitive capabilities (compute, memory, reasoning, external call, observability), (P6) subsumption asymmetry showing structural governance strictly subsumes content-level filtering, and (P7) semantic transparency: on all executions where governance permits, the governed interpretation is observationally equivalent (modulo governance-only events) to the ungoverned interpretation. Together, these results show that governance and computational expressivity are orthogonal dimensions: governance constrains the effect boundary of programs while remaining semantically transparent to internal computation.

Algebraic Semantics of Governed Execution: Monoidal Categories, Effect Algebras, and Coterminous Boundaries

arXiv:2605.01032v2 Announce Type: new Abstract: We present an algebraic semantics for governed execution in which governance is axiomatized, compositional, and coterminous with expressibility. The framework, mechanized in 32 Rocq modules (~12,000 lines, 454 theorems, 0 admitted), is built on interaction trees and parameterized coinduction. A three-axiom GovernanceAlgebra record (safety, transparency, properness) induces a symmetric monoidal category with verified pentagon, triangle, and hexagon coherence, where every tensor composition preserves governance. An algebraic effect system constrains the handler algebra so that only governance-preserving handlers can be constructed in the safe fragment; programs in the empty capability set provably emit only observability directives. Capability-indexed composition bundles programs with machine-checked capability bounds, and a dual guarantee theorem establishes that within_caps and gov_safe hold simultaneously under all composition operators. The capstone result is the coterminous boundary: within our formal model, every program expressible via the four primitive morphism constructors is governed under interpretation, and every governed program is the image of such a program. Turing completeness is preserved inside governance; unmediated I/O is excluded from the governed fragment. Governance denial is modeled as safe coinductive divergence. The governance algebra is parametric: any system instantiating the three axioms inherits all derived properties, including convergence, compositional closure, and goal preservation. Extracted OCaml runs as a NIF in the BEAM runtime, with property-based testing (70,000+ random inputs, zero disagreements) confirming behavioral equivalence between the specification and the runtime interpreter.

CISA urges critical infrastructure firms to ‘fortify’ before it’s too late | Cybersecurity Dive

As concerns mount about potential cyber sabotage by the Chinese government, the U.S. is warning operators to practice maintaining services in a degraded state.

US government warns of severe CopyFail bug affecting major versions of Linux

Federal authorities have issued a warning regarding a critical vulnerability known as 'CopyFail' that impacts several versions of the Linux operating system.

Attackers are cashing in on fresh 'CopyFail' Linux flaw

Researchers dropped a reliable root exploit and it didn’t sit idle for long CISA is warning that a newly-disclosed Linux kernel bug dubbed "CopyFail" is already being exploited, just days after researchers dropped a working root-level exploit.…

Cisco Unveils Open-Source Toolkit to Bolster AI Model Security and Provenance

Cisco has unveiled the open-source Model Provenance Kit to bolster AI model security by tracking their origin and integrity.

Adoption Was AI's Problem. Now It's a Business Model. | PYMNTS.com

Most enterprise artificial intelligence projects don’t fail because the technology doesn’t work. They fail because the technology never makes it into the

Enterprise AI Works in Notebooks, Not in Production — Inside the Deployment Breakdown - KoreaTechDesk | Korean Startup and Technology News

When a Working Model Stops Working — An enterprise AI model can pass internal testing, meet accuracy benchmarks, and still fail when deployed into a real system. This pattern is becoming common across enterprise AI. The issue is not whether models can be built.

ClinicBot: A Guideline-Grounded Clinical Chatbot with Prioritized Evidence RAG and Verifiable Citations

arXiv:2605.00846v1 Announce Type: new Abstract: Clinical diagnosis requires answers that are accurate, verifiable, and explicitly grounded in official guidelines. While large language models excel at natural language processing, their tendency to hallucinate undermines their utility in high-stakes medical contexts where precision is essential. Existing retrieval-augmented generation (RAG) systems treat all evidence equally, producing noisy context and generic answers misaligned with clinical practice. We present ClinicBot, an AI system that translates guideline recommendations into trustworthy clinical support through three key advances: (1) structured extraction of clinical guidelines into semantic units (recommendations, tables, definitions, narrative) with explicit provenance, (2) evidence prioritization that ranks content by clinical significance and guideline structure rather than textual similarity, and (3) a web-based interface that presents concise, actionable answers with verifiable evidence. We will demonstrate ClinicBot using diabetes questions from real patients and an additional diabetes risk assessment tool that is faithful to the American Diabetes Association (ADA) Standards of Care in Diabetes (2025). The demonstration will illustrate how semantic knowledge extraction and hierarchical evidence ranking can reliably operate in a multi-agent setting to process complex clinical guidelines at scale.

PERSA: Reinforcement Learning for Professor-Style Personalized Feedback with LLMs

arXiv:2605.01123v1 Announce Type: new Abstract: Large language models (LLMs) can provide automated feedback in educational settings, but aligning an LLMs style with a specific instructors tone while maintaining diagnostic correctness remains challenging. We ask how can we update an LLM for automated feedback generation to align with a target instructors style without sacrificing core knowledge? We study how Reinforcement Learning from Human Feedback (RLHF) can adapt a transformer-based LLM to generate programming feedback that matches a professors grading voice. We introduce PERSA, an RLHF pipeline that combines supervised fine-tuning on professor demonstrations, reward modeling from pairwise preferences, and Proximal Policy Optimization (PPO), while deliberately constraining learning to style-bearing components. Motivated by analyses of transformer internals, PERSA applies parameter efficient fine-tuning. It updates only the top transformer blocks and their feed-forward projections, minimizing global parameter drift while increasing stylistic controllability. We evaluate our proposed approach on three code-feedback benchmarks (APPS, PyFiXV, and CodeReviewQA) using complementary metrics for style alignment and fidelity. Across both Llama-3 and Gemma-2 backbones, PERSA delivers the strongest professor-style transfer while retaining correctness, for example on APPS, it boosts Style Alignment Score (SAC) to 96.2% (from 34.8% for Base) with Correctness Accuracy (CA) up to 100% on Llama-3, and Gemma-2. Overall, PERSA offers a practical route to personalized educational feedback by aligning both what it says (content correctness) and, crucially, how it says it (instructor-like tone and structure).

A Knowledge-Driven LLM-Based Decision-Support System for Explainable Defect Analysis and Mitigation Guidance in Laser Powder Bed Fusion

arXiv:2605.01100v1 Announce Type: new Abstract: This work presents a knowledge-driven decision-support system that integrates structured defect knowledge with LLM-based reasoning to provide explainable defect diagnosis and mitigation guidance in manufacturing, using LPBF as a representative, safety-critical case study. The proposed ontology-integrated LLM-based decision support system for LPBF defect analysis and mitigation guidance is built on a knowledge base containing 27 known LPBF defect types organized into hierarchical categories and causal relationships. The developed system supports fuzzy natural language queries for systematic knowledge retrieval, literature-supported explanation of defects, and guidance on defect causes and mitigation strategies derived from encoded process knowledge. Furthermore, a multimodal image-assessment module based on foundation models enables descriptor-guided interpretation of representative microscopic defect images through semantic alignment scoring. The proposed framework was evaluated through qualitative comparisons with general-purpose vision-language models, an ablation study, and an inter-rater reliability analysis. Evaluation on the literature-derived dataset showed that the fully integrated configuration outperformed the other three evaluated system configurations, achieving a macro-average F1 score of 0.808. Additionally, inter-rater reliability analysis using Cohen's kappa indicated substantial agreement between the model outputs and the literature-derived reference labels. These findings suggest that ontology-guided knowledge representation can improve the consistency, interpretability, and practical usefulness of LLM-assisted LPBF defect analysis.

GPT-5.5 Instant shows you what it remembered — just not all of it

OpenAI updated the default model for ChatGPT to its new GPT-5.5 Instant, along with a new memory capability that finally shows which context shaped responses — at least some of them.  This limitation signals that models are starting to create a second, incomplete memory observability layer that could conflict with existing audit systems and agent logs.  GPT-5.5 Instant replaces GPT-5.3 Instant as the default ChatGPT model and is a version of its new flagship GPT-5.5 LLM. It’s supposed to be more dependable, accurate and smarter than 5.3.  But it’s the introduction of memory sources, which will be enabled across all models in the platform, that could help enterprises in their projects.  “When a response is personalized, you can see what context was used, such as saved memories or past chats, and delete or correct it if something is outdated or no longer relevant,” OpenAI said in a blog post.  When a user asks ChatGPT something, users can tap the sources button (at the bottom of the response) to see which files or past chats the model tapped to find the answer. Users also have full control over the sources models can cite, and these sources will not be shared if the conversation is sent to others.  The company said memory sources should make it easier to personalize model responses. Still, OpenAI admitted that the models “may not show every factor that shaped an answer” and promised to make the capability more comprehensive over time.  What this means is that memory sources offer a semblance of observability in ChatGPT answers, but not full auditability yet.  Competing memory systems  Enterprises have a system in place to solve part of the memory and context problem with models and agents. Models are exposed to context through retrieval-augmented generation (RAG) pipelines; whatever the agent fetches from the vector databases is logged, and the agent's state is stored in a memory layer. All of this is tracked in application logs, usually in an orchestration or management layer with built-in observability. Ideally, this allows teams to trace failure back through the stack. The current system is imperfect; sometimes, it's not easy to trace failure points, but it’s at least internally consistent. For enterprises using ChatGPT, whether the default GPT-5.5 Instant or their model of choice, that’s no longer the case. The model surfaces its own version with memory sources that are wholly separate from existing retrieval logs — in short, a model-reported context. A problem arises if these cannot be reconciled reliably. And because memory sources only give users part of the picture — it’s unclear what ChatGPT’s limit on citing memory sources is — it becomes even harder to match what GPT-5.5 Instant said it tapped to what it actually did in the production environment. This situation creates a new failure mode: A competing context log. If something seems wrong, it can create inconsistencies that enterprises have to deal with. Malcolm Harkins, chief trust and security officer at HiddenLayer, told VentureBeat that memory sources "look like a pragmatic middle ground " in offering some transparency, but it's still not easy to see its value. "For enterprises, it's directionally useful but insufficient on its own," Harkins said. "Real value will depend on how it integrates with security, governance, access controls and audit systems." A more capable default model  However, GPT-5.5 Instant handles memory, and OpenAI calls it an improvement over GPT-5.3 Instant.  Internal evaluations showed GPT-5.5 Instant returned 52.5% fewer hallucinated claims than the previous default model, especially for high-stakes domains such as medicine, law, and finance. Inaccurate claims fell by 37.3% on challenging conversations. The company said the model improved on photo analysis and image uploads, answering STEM questions and knowing when to tap its own knowledge base or use web search.  Peter Gostev, AI capability at independent model evaluator Arena, explained to VentureBeat in an email that the key result to watch about GPT-5.5 Instant is how it performs on the overall text rankings, especially because its predecessor did not have a strong showing.  “Since GPT-4o, the strongest-performing OpenAI chat model on the Arena has been GPT-5.2-Chat, which still ranks 12th on the Overall Text Arena months after release," Gostev said. Notably, users preferred it even over the higher-reasoning GPT-5.2-High variant, which is currently ranked 52nd on the Arena. “By comparison, GPT-5.3-Chat, the previous default model in ChatGPT, was significantly less competitive, ranking 44th overall, 32 places below GPT-5.2-Chat.” What enterprises need to do about memory sources Organizations that rely on ChatGPT for some tasks will need to formalize how memory works for their stack. Memory sources are not limited to GPT-5.5 Instant; it is enabled for all models on the ChatGPT platform.  To address the problem of competing memory sources, enterprises have to audit their memory management. Model-reported context could overlap or contradict these logs, so it’s best to define a clear source of truth. In the event of a failure, administrators know which log to believe.  It would also be a good idea to decide whether or not to expose memory sources to users. ChatGPT only shows a select number of chats or files it used to complete a request. Some users may find more transparency trustworthy.  Ultimately, the number one thing for enterprises to remember about memory sources is that what the model reports as its context is not the full picture for auditing. It’s a form of observability, but it cannot withstand a full examination.

Behind the AI in the Newsroom: The Washington Post’s Vineet Khosla

In this episode of Me, Myself, and AI, host Sam Ransbotham speaks with Vineet Khosla, CTO of The Washington Post, about how AI is reshaping the way news is produced, delivered, and consumed. Vineet argues that journalism itself isn’t broken — but the formats people use to consume news are rapidly evolving, especially as audiences […]

To Use AI as Dice of Possibilities with Timing Computation

arXiv:2605.01134v1 Announce Type: new Abstract: The dominant noun-based modeling paradigm has fundamentally constrained AI development, precluding any adequate representation of the future as an open temporal dimension. This paper introduces a verb-based paradigm, together with precise definitions of \emph{timing computation} and \emph{possibility}, that enables AI to function as an effective instrument for realizing the grammar of our thought. Applied to longitudinal EHR data from 3,276 breast cancer patients, the framework empirically demonstrates: (1) automatic discovery of clinically significant patient trajectories, and (2) counterfactual timing deduction. Both results are purely data-driven, require no prior domain knowledge, and, to our knowledge, represent the first such demonstrations in the machine learning literature.

Singapore central bank pilots cross-bank AI to detect scams earlier

The Monetary Authority of Singapore is collaborating with five banks and government agencies to test AI and machine learning for pre-emptive scam detection using pooled cross-bank transaction data.

Deco: Extending Personal Physical Objects into Pervasive AI Companion through a Dual-Embodiment Framework

arXiv:2605.03882v1 Announce Type: cross Abstract: Individuals frequently form deep attachments to physical objects (e.g., plush toys) that usually cannot sense or respond to their emotions. While AI companions offer responsiveness and personalization, they exist independently of these physical objects and lack an ongoing connection to them. To bridge this gap, we conducted a formative study (N=9) to explore how digital agents could inherit and extend the emotional bond, deriving four design principles (Faithful Identity, Calibrated Agency, Ambient Presence, and Reciprocal Memory). We then present the Dual-Embodiment Companion Framework, instantiated as Deco, a mobile system integrating multimodal Large Language Models (LLMs) and Augmented Reality to create synchronized digital embodiments of users' physical companions. A within-subjects study (N=25) showed Deco significantly outperformed a personalized LLM-empowered digital companion baseline on perceived companionship, emotional bond, and design-principle scales (all p<0.01). A seven-day field deployment (N=17) showed sustained engagement, subjective well-being improvement (p=.040), and three key relational patterns: digital activities retroactively vitalized physical objects, bond deepening was driven by emotional engagement depth rather than interaction frequency, and users sustained bonds while actively navigating digital companions' AI nature. This work highlights a promising alternative for designing digital companions: moving from creating new relationships to dual embodiment, where digital agents seamlessly extend the emotional history of physical objects.

SP Energy Networks and Keen AI launch digital tool to cut UK grid connection wait times

ScottishPower transmission and distribution subsidiary SP Energy Networks has partnered with UK AI company Keen AI to deploy an AI-powered tool that provides greater visibility into transmission grid connection options for energy developers. – Sebastian Moss The tool, dubbed IConn, works by digitalizing the transmission network into a unified view of existing, contracted, and planned […]

Penske Launches AI-Driven Platform for Real-Time Supply Chain Visibility and Efficiency

Penske Logistics has unveiled Supply Chain Insight, a cloud-native platform offering real-time visibility and AI-driven decision-making for supply chain performance.

How AI helped this 27-year-old boost his investments by $75,000 | Reuters

Alex Caswell, CEO of Wealth Script Advisors, says he understands why more young people are turning to AI : it’s cheaper than hiring a financial advisor and offers quick answers as they try to manage their money.

Geographic Variation in Stack Overflow Code Quality: Evidence from a Cross-Regional Study of Coding Practices

arXiv:2605.03670v1 Announce Type: cross Abstract: Developers frequently reuse Stack Overflow code snippets, yet the quality of these snippets remains unevenly understood, particularly across programming languages and geographic contexts. This study investigates code quality in Stack Overflow answers from contributors located in the United States, focusing on SQL, JavaScript, Python, Ruby, and Java snippets. We evaluate four quality dimensions: reliability, readability, performance, and security. Using language-specific linting and static analysis tools, we quantify violations across states and cities, compute violation densities to enable fair regional comparison, and examine relationships between code quality and state-level diversity indicators. We further conduct inductive content analysis on code snippets from California, Utah, and North Dakota to identify qualitative patterns in code quality violations. Results show that readability violations are the most prevalent across all languages, followed by reliability, performance, and security. Common issues include improper whitespace, inconsistent formatting, program-flow errors, inefficient resource use, unsanitised inputs, and insecure dynamic evaluation. Regional analysis indicates that major technology hubs produce more parsable snippets but do not necessarily exhibit higher violation densities. States with broader access to computing devices, Internet subscriptions, higher income, and more equitable wealth distribution tend to show fewer code quality violations. Qualitative findings suggest that established technology regions often produce more complex violations, while less mature technology regions display more fundamental errors. These findings highlight the socio-technical nature of code quality in community question-answering platforms and suggest that developers should exercise caution when reusing online code snippets.

How foundries are shaping the next era of enterprise AI | TechRadar

The architecture fixing enterprise AI fragmentation

AI as Advisor: Anthropic Flags Growing Use of Claude for Personal Guidance

Anthropic study shows users increasingly rely on Claude for personal guidance, with 75% of advice queries focused on health, career, relationships, and finance

MoveOD: Synthesizing Origin-Destination Commute Distribution from U.S. Census Data

arXiv:2510.18858v2 Announce Type: replace Abstract: High-resolution origin-destination (OD) tables are essential for a wide spectrum of transportation applications, from modeling traffic and signal timing optimization to congestion pricing and vehicle routing. However, outside a handful of data rich cities, such data is rarely available. We introduce MOVEOD, an open-source pipeline that synthesizes public data into commuter OD flows with fine-grained spatial and temporal departure times for any county in the United States. MOVEOD combines five open data sources: American Community Survey (ACS) departure time and travel time distributions, Longitudinal Employer-Household Dynamics (LODES) residence-to-workplace flows, county geometries, road network information from OpenStreetMap (OSM), and building footprints from OSM and Microsoft, into a single OD dataset. We use a constrained sampling and integer-programming method to reconcile the OD dataset with data from ACS and LODES. Our approach involves: (1) matching commuter totals per origin zone, (2) aligning workplace destinations with employment distributions, and (3) calibrating travel durations to ACS-reported commute times. This ensures the OD data accurately reflects commuting patterns. We demonstrate the framework on Hamilton County, Tennessee, where we generate roughly 150,000 synthetic trips in minutes, which we feed into a benchmark suite of classical and learning-based vehicle-routing algorithms. The MOVEOD pipeline is an end-to-end automated system, enabling users to easily apply it across the United States by giving only a county and a year; and it can be adapted to other countries with comparable census datasets. The source code and a lightweight browser interface are publicly available.

Bosses Want You to Use AI but They're Not Setting a Good Example, Study Says - CNET

Executives have been pushing AI for years. But leadership-led workplace culture may be preventing effective adoption, Microsoft found.

CEOs and Boards Are Aligned on AI in Theory, but Divided in Practice

This article examines gaps between CEOs and boards regarding AI governance, ROI expectations, implementation speed, and board-level AI literacy.

When everyone has AI and the company still learns nothing

Are people using AI, or is the organization learning from it? What changed because we spent those tokens? And who moves discoveries from individuals to teams to organizational capabilities?

The enterprise productivity fix hiding in the pit lane - SiliconANGLE

F1 offers an extreme proxy for enterprise performance — every decision is data-driven, every week demands measurable progress and the cost of poor knowledge flow is visible in race results, Boyagi noted. The same systemic barriers blocking software teams, including unclear priorities and AI that ...

Track AI ROI in Real Time

A guide on connecting spend, infrastructure, and model performance to help teams troubleshoot cost spikes and make smarter decisions about AI ROI.

4 ways HR uses AI in hiring, according to talent professionals | HR Dive

Inefficient implementation, however, may be cutting into employers’ ROI.

Political Economy Constraints Will Likely Shape the Regulatory Future of AI Adoption

Professional groups with significant political influence are positioned to shape AI regulation to protect their interests. The trajectory of AI deployment will be determined as much by political contestation as by technical capability.

Protesters push Portland to investigate firm that appears to supply drone tech to Israel

Sightline Intelligence sent AI-supported tool to company that provides drones to Israeli military, research group says Anti-war activists in Portland, Oregon, are pushing city authorities to ensure no local resources, tax breaks or investments support a local company that appears to be supplying artificial intelligence software to the Israeli military. The company, Sightline Intelligence, manufactures AI-supported video technology that is used in drones to interpret target movements and make quick decisions based on the perceived threat level. Cargo documents appear to show Sightline has shipped its technology to Elbit Systems, an Israeli arms manufacturer that provides drones to that country’s military and exports to others. The activists argue that such sales violate the UN’s arms agreements. Continue reading...

The AI Regulation Regime Shift: Trump Administration’s Sudden U-Turn on Frontier Models

Daily Market Read I | Category: regime read | Date: May 5, 2026

NATO needs policies, standards for sharing AI-enhanced geospatial intel: Official - Breaking Defense

"The path to AI enabled, allied intelligence advantage runs primarily through governance, not necessarily through additional capability," said UK Royal Marine Maj. Gen. Paul Lynch, who directs NATO intelligence policy.

Google, Microsoft, and xAI agree to pre-release government AI model evaluations as Mythos crisis forces oversight expansion

Five frontier AI labs now submit models for US government evaluation. The voluntary programme has no statutory authority but covers every major AI developer after the Mythos crisis.

Microsoft, Google, and xAI Grant US Early Access to AI Models for Security Testing – Firstpost

Google, Microsoft, and xAI will give the US government early access to their AI models to assess national security risks

US government expands vetting of frontier AI models for security risks - POLITICO

The Commerce Department’s Center for AI Standards and Innovation will conduct safety testing of new AI systems before they are released publicly.

Meta Hit With Massive Lawsuit—Publishers Say AI Was Trained on “Stolen” Books

Publishers have filed a major lawsuit against Meta, alleging that their AI models were trained on copyrighted books without authorization.

The Download: inside the Musk v. Altman trial, and AI for democracy

This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. Week one of the Musk v. Altman trial: what it was like in the room Two of the most powerful figures in AI—Sam Altman and Elon Musk—are in the middle of…

What Smart People Are Saying About Trump Weighing AI Oversight - Business Insider

Tech analysts said they worry oversight could slow innovation, especially as the US tries to keep pace with China.

Apple agrees to pay iPhone owners $250 million for not delivering AI Siri

Apple has reached a settlement in a class-action lawsuit regarding unfulfilled promises about Siri's AI capabilities.

As false information created by generative artificial intelligence (AI) leads to actual damage, laws.. - MK

As false information created by generative artificial intelligence (AI) leads to actual damage, lawsuits over related responsibilities are continuing. In Canada, Google AI incorrectly identified a fam..