Thu 21 May 2026
Daily Brief — Curated and contextualised by Best Practice AI
Big Tech Borrows, Trump Intervenes, and Kyndryl Cuts Jobs
TL;DR Tech companies are borrowing heavily to fund AI infrastructure, straining the investment-grade bond market. Trump plans an executive order for AI oversight, marking a shift in US policy. OpenAI is preparing for a $1 trillion IPO. Nvidia's profits have tripled, driven by AI demand, while Kyndryl plans workforce cuts to save $500 million.
The stories that matter most
Selected and contextualised by the Best Practice AI team
The Economics of AI Inference: Inflation Dynamics, Welfare Costs, and Optimal Monetary Policy under the Inference-Cost Phillips Curve
arXiv:2605.20281v1 Announce Type: new Abstract: We develop a unified microeconomic and monetary theory of artificial intelligence inference costs and their pass-through to inflation, welfare, and optimal monetary policy. We introduce the Inference-Cost Phillips Curve (ICPC), an augmented New Keynesian Phillips curve in which firm-level marginal costs of producing differentiated goods include a non-trivial AI inference component lambda-bar, and prove a closed-form structural slope kappa*_inf = lambda-bar * kappa, where kappa is the standard Calvo-Yun slope. We derive a welfare-relevant Hicks-Kaldor decomposition of consumer welfare under inference-cost shocks, prove a generalized Taylor principle for the inference-augmented economy, and characterize the optimal monetary policy response coefficient psi*_inf = (1 + phi*rho) * lambda-bar * kappa under commitment. A second-order welfare loss formula closes the model in closed form. We confront the theory with U.S. monthly data 2022:M01-2026:M04 using a two-step GMM estimator with Newey-West HAC standard errors and Hansen J-test, recovering an empirical slope kappa-hat_inf = 0.087 (HAC s.e. 0.021) which lies within one standard error of the structural prediction. A scaling regression over 50 rolling-window subwindows yields b-hat = 0.987 (R^2 = 0.998), consistent with a near-unit-elasticity pass-through. A G7 reduced-form panel with Driscoll-Kraay HAC standard errors yields b-hat^G7 = 0.094 (s.e. 0.026), and a Wald test fails to reject cross-country homogeneity (p = 0.78). The framework provides a single equilibrium scaffold for the joint study of AI inference cost dynamics, monetary policy under generative-AI shocks, and the welfare cost of inference-driven inflation.
The Economics of Model Collapse: Equilibrium, Welfare, and Optimal Provenance Subsidies in Synthetic Data Markets
arXiv:2605.20279v1 Announce Type: new Abstract: Generative artificial intelligence is rapidly transforming the supply side of training data: an increasing share of new tokens, images, and structured records is produced by previous-generation models rather than by human originators. Recursive training on such synthetic content induces a measurable and often irreversible loss of distributional fidelity, a phenomenon known as model collapse. We develop the first unified microeconomic theory of synthetic data markets under model collapse. We introduce the Synthetic Data Contamination Equilibrium (SDCE), prove existence and generic uniqueness, derive a welfare decomposition W = W_prod + W_cons - L_coll - L_info, establish a Wasserstein-gradient-flow mean-field collapse limit, prove an impossibility of information-constrained implementation, and obtain closed-form expressions for the welfare-maximizing provenance subsidy s* = KL(q||p)/(2 kappa) and the welfare-maximizing watermark strength w* = (1 - psi) KL(q||p)/(2 kappa psi). We prove an information-theoretic Cramer-Rao lower bound on any provenance estimator using only producer-side observations and show that the Provenance-Market Iterative Retraining (PMIR) algorithm attains this bound up to constants while converging to an epsilon-SDCE in O(epsilon^-2 log T) iterations. A reduced-form OLS estimation on a C4-synthetic benchmark over ten retraining generations yields a collapse-rate coefficient b-hat = 0.181 (HAC s.e. 0.024), within one standard error of the structural prediction 0.183. Calibrated experiments raise generation-ten model quality by 23.1 percent over the unregulated benchmark while lowering the 2-Wasserstein drift on a held-out diversity probe from 0.318 to 0.142. Scaling experiments over generations t in {1,...,10} recover a logarithmic-in-t collapse law log Q_t = log Q_0 - 0.183 t rho^2 with R^2 = 0.962.
UN Draft Treaty Aims to Boost Nations’ Rights to Tax Tech Giants
Countries at the United Nations are rewriting international tax rules in an effort to be able to tax technology giants like Alphabet and Amazon based on where their users are located rather than where they’re headquartered.
'Workforce rebalancing' comes for Kyndryl, and delivery teams are in the firing line
Kyndryl is targeting up to $500 million in savings through workforce rebalancing and the adoption of agentic AI.
Swiss giant battery developer taps UK tech to feed AI power boom
World’s largest vanadium flow battery project selects Invinity Energy Systems to meet data centre energy demands
Economics & Markets
The impossible maths of the AI boom - Klement on Investing
A couple of months ago I wrote in my regular Reuters column about LLMs and why I think they do not have a sustainable business model. In a second piece, which is based on an extensive research note I wrote I turn my attention to hyperscalers and the datacentre investment boom.
Why AI's Productivity Gains Are Showing Up In Some Earnings And Not Others
The make-versus-buy decision is one of the more consequential strategic choices playing out in real time. Companies that are building proprietary AI capabilities are making a bet that the competitive moat from owning the model outweighs the capital cost of building it.
AI Debt Boom Sparks Warnings
Danielle Poli, Managing Director and Co-Portfolio Manager of Global Credit at Oaktree Capital Management joined Bloomberg Open Interest to break down the hidden risks behind the AI boom, why higher interest rates may be here longer than markets expect, and where investors are still finding 8–10% yields. She warns that massive AI debt issuance could create future cracks in credit markets, especially in private credit and leveraged loans. (Source: Bloomberg)
AI Trade Still in 'Very Early Innings,' PNC's Agati Says
PNC Asset Management Chief Investment Officer Amanda Agati sees "a little bit of steam coming out of the market rally sails." She speaks on "Bloomberg Open Interest." (Source: Bloomberg)
OpenAI Prepares to File for an I.P.O. in Coming Weeks
OpenAI would be one of the most highly anticipated potential initial public offerings, in what is set to be a major year of I.P.O.s for Silicon Valley.
Polymarket opens public price discovery on private startups with Nasdaq Private Market deal - Startup Fortune
Polymarket has launched prediction markets for private-company milestones via a deal with Nasdaq Private Market, creating public price signals for startup
Big Tech software era is over, says top investor James Anderson
Former Baillie Gifford fund manager says spoils of AI war will flow to hardware suppliers
AI boom reshuffles global stock market pecking order as South Korea and Taiwan surge
A global reshuffling in stock-market hierarchy is underway. AI is propelling Taiwan and South Korea past a couple long-established Western countries.
SpaceX prepares to make S-1 public, turning private-valued Starlink into a market benchmark - Startup Fortune
SpaceX's anticipated public S-1 will for the first time show how much revenue and profit come from Starlink versus launch services, set a new valuation
The benefits of an investment bubble
UK telecoms sector shows how splurge of money can destroy shareholder value but deliver consumer gains
Pension funds face growing AI concentration risk in venture capital as valuations soar - Pensions & Investments
Artificial intelligence companies’ escalating market values accounts for an increasing portion of the U.S. venture capital market.
When AI giants go public, will ordinary investors know if they are along for the ride?
Large index-tracking funds could soon gain automatic exposure to AI giants as companies such as OpenAI edge toward public markets.
Tracker of AI Company Spend Vs Revenue Shows That Everyone's Broke Except Nvidia - Gadget Review
AI companies have spent $1.4 trillion on infrastructure but earned only half that in revenue, while Nvidia alone profits $253 billion from the boom.
Nvidia Stock Heads Into Earnings With Biggest Short Position In S&P 500 — HSBC Sees ‘Beat And Raise’ Quarter
The options market implies a potential $3.5 billion one-day swing for short sellers after earnings.
AI Needs Power. This Soaring ETF Has the Right Stuff. | The Motley Fool
The Defiance AI & Power Infrastructure ETF puts investors front and center with the AI hardware and power booms.
The Economics of Model Collapse: Equilibrium, Welfare, and Optimal Provenance Subsidies in Synthetic Data Markets
arXiv:2605.20279v1 Announce Type: new Abstract: Generative artificial intelligence is rapidly transforming the supply side of training data: an increasing share of new tokens, images, and structured records is produced by previous-generation models rather than by human originators. Recursive training on such synthetic content induces a measurable and often irreversible loss of distributional fidelity, a phenomenon known as model collapse. We develop the first unified microeconomic theory of synthetic data markets under model collapse. We introduce the Synthetic Data Contamination Equilibrium (SDCE), prove existence and generic uniqueness, derive a welfare decomposition W = W_prod + W_cons - L_coll - L_info, establish a Wasserstein-gradient-flow mean-field collapse limit, prove an impossibility of information-constrained implementation, and obtain closed-form expressions for the welfare-maximizing provenance subsidy s* = KL(q||p)/(2 kappa) and the welfare-maximizing watermark strength w* = (1 - psi) KL(q||p)/(2 kappa psi). We prove an information-theoretic Cramer-Rao lower bound on any provenance estimator using only producer-side observations and show that the Provenance-Market Iterative Retraining (PMIR) algorithm attains this bound up to constants while converging to an epsilon-SDCE in O(epsilon^-2 log T) iterations. A reduced-form OLS estimation on a C4-synthetic benchmark over ten retraining generations yields a collapse-rate coefficient b-hat = 0.181 (HAC s.e. 0.024), within one standard error of the structural prediction 0.183. Calibrated experiments raise generation-ten model quality by 23.1 percent over the unregulated benchmark while lowering the 2-Wasserstein drift on a held-out diversity probe from 0.318 to 0.142. Scaling experiments over generations t in {1,...,10} recover a logarithmic-in-t collapse law log Q_t = log Q_0 - 0.183 t rho^2 with R^2 = 0.962.
The Economics of AI Inference: Inflation Dynamics, Welfare Costs, and Optimal Monetary Policy under the Inference-Cost Phillips Curve
arXiv:2605.20281v1 Announce Type: new Abstract: We develop a unified microeconomic and monetary theory of artificial intelligence inference costs and their pass-through to inflation, welfare, and optimal monetary policy. We introduce the Inference-Cost Phillips Curve (ICPC), an augmented New Keynesian Phillips curve in which firm-level marginal costs of producing differentiated goods include a non-trivial AI inference component lambda-bar, and prove a closed-form structural slope kappa*_inf = lambda-bar * kappa, where kappa is the standard Calvo-Yun slope. We derive a welfare-relevant Hicks-Kaldor decomposition of consumer welfare under inference-cost shocks, prove a generalized Taylor principle for the inference-augmented economy, and characterize the optimal monetary policy response coefficient psi*_inf = (1 + phi*rho) * lambda-bar * kappa under commitment. A second-order welfare loss formula closes the model in closed form. We confront the theory with U.S. monthly data 2022:M01-2026:M04 using a two-step GMM estimator with Newey-West HAC standard errors and Hansen J-test, recovering an empirical slope kappa-hat_inf = 0.087 (HAC s.e. 0.021) which lies within one standard error of the structural prediction. A scaling regression over 50 rolling-window subwindows yields b-hat = 0.987 (R^2 = 0.998), consistent with a near-unit-elasticity pass-through. A G7 reduced-form panel with Driscoll-Kraay HAC standard errors yields b-hat^G7 = 0.094 (s.e. 0.026), and a Wald test fails to reject cross-country homogeneity (p = 0.78). The framework provides a single equilibrium scaffold for the joint study of AI inference cost dynamics, monetary policy under generative-AI shocks, and the welfare cost of inference-driven inflation.
Is Nvidia too big to fail?
‘You’re clearly at the centre of everything’
Nvidia's Huang Is 'Dunking' on Competition, Luria Says
Gil Luria, DA Davidson head of technology research, says Nvidia CEO Jensen Huang is "dunking" on all the competition. Joe Kaiser of Switchyard Partners also reacts to Nvidia's earnings and SpaceX going public. They are on "Bloomberg Open Interest." (Source: Bloomberg)
ParlayMarket: Automated Market Making for Parlay-style Joint Contracts
arXiv:2603.22596v2 Announce Type: replace-cross Abstract: Prediction markets are powerful mechanisms for information aggregation, but existing designs are optimized for single-event contracts. In practice, traders frequently express beliefs about joint outcomes - through parlays in sports, conditional forecasts across related events, or scenario bets in financial markets. Current platforms either prohibit such trades or rely on ad hoc mechanisms that ignore correlation structure, resulting in inefficient prices and fragmented liquidity. We introduce ParlayMarket, the first automated market-making design that supports parlay-style joint contracts within a unified liquidity pool while maintaining coherent pricing across base markets and their combinations. Our main result is a convergence characterization of the resulting system. Under repeated trading, the AMM dynamics converge to a unique fixed point corresponding to the best approximation to the true joint distribution within the model class. We show that (i) parameter error remains bounded at stationarity due to a balance between signal and noise in trade-induced updates, and (ii) pricing error and monetary loss scale with this parameter error, implying that aggregate market-maker loss remains controlled and grows at most quadratically in the number of base markets. These results establish explicit limits on the information-retrieval error achievable through the trading interface. Importantly, parlay trades play a structural role in this convergence: by providing direct constraints on joint outcomes, they improve identifiability of dependence structure and reduce steady-state error relative to markets that rely only on marginal trades. Empirically, we show both in controlled simulations and in replay on historical Kalshi parlay data that this design achieves the intended scaling while remaining effective in realistic market settings.
Allbirds’ 600% stock surge says a lot about how ‘AI washing’ became the new ‘greenwashing’
In the early 2000s, corporate sustainability faced a similar credibility crisis — every company measured sustainability differently.
India’s Competition Regulator Warns of AI-Driven Market Risks | PYMNTS.com
Experts added that regulators worldwide ... how dominant firms may use AI-driven systems to reinforce market advantages, particularly in digital markets where algorithms play a significant role in shaping consumer choices and competition dynamics. ... Generative AI and Competitive Disruption: Increasingly Relevant for Merger Analysis...
Index: A Platform for Content Owners
Index is a platform designed to help content owners track how AI agents use their work, signaling a shift toward agent-readable web infrastructure and content compensation.
Google Cloud suspended major customer Railway.com without cause, causing outage
Railway.com, an automated code deployment outfit, reports a service outage after Google Cloud suspended their account without cause.
With ecosystem advantage, can Google challenge Meta in smart glasses market | Tech News - Business Standard
Smart Glasses Market: AI-led smart glasses are entering a high-growth phase globally, but Meta's early dominance and Google's ecosystem-driven strategy could shape how competition and adoption evolve
Can AI make the public sector more efficient?
Productivity gains may be cancelled out by the public’s own use of the technology when interacting with authorities
AI didn’t kill brand consistency — it made it mission-critical
Presented by Design.com Generative AI has made design radically more accessible. A founder can now create a logo, launch a website, build social campaigns, generate presentations, and produce marketing collateral in a single afternoon — work that once required agencies, freelancers, or internal creative teams. But as design generation becomes easier, maintaining a recognizable identity becomes harder. The problem is no longer whether businesses can create content. It’s whether all of that content still feels like it comes from the same company. That shift matters most for emerging businesses. Established enterprises already have brand governance systems, design teams, and years of customer recognition reinforcing their identity. Small businesses and solo founders often have none of those advantages. Their brand is built almost entirely through digital touchpoints — websites, presentations, social posts, ads, emails, and customer interactions that may be created across multiple tools and platforms. In the AI era, inconsistency scales just as fast as creativity. AI has turned branding into a systems problem The biggest risk with AI-generated design is not necessarily poor-quality output. In many cases, individual assets look polished on their own. The problem is fragmentation. A logo generated in one tool may not align with the visual language of a website created elsewhere. Marketing graphics evolve independently from presentation templates. Messaging shifts between channels. Colors, typography, layouts, and tone gradually drift as more assets are produced. Over time, the business stops presenting a coherent identity. Consumers increasingly encounter brands through dozens of micro-interactions rather than a single destination. A customer may discover a company through social media, visit its website, receive an email campaign, then view a proposal or presentation later. If those experiences feel disconnected, credibility erodes quickly — particularly for younger companies still trying to establish trust and legitimacy in crowded digital markets. That changes the role of design itself. Instead of treating design as a series of one-off deliverables, businesses increasingly need connected brand systems that carry identity consistently across every asset they create. Why static style guides are no longer enough Traditional brand style guides were built for slower creative cycles. Teams manually referenced approved logos, fonts, colors, and tone guidelines while producing a relatively limited number of assets. AI changes both the speed and scale of content generation. When businesses are producing dozens — or hundreds — of design variations across channels, consistency cannot rely entirely on humans manually enforcing brand rules after the fact. Brand governance increasingly needs to become embedded directly into the creation process itself, allowing brand rules and visual systems to persist across every generated asset. That is where platforms like Design.com are trying to evolve beyond standalone design generation tools. Rather than treating a logo, website, presentation, and marketing assets as separate projects, the platform carries core branding decisions — visual identity, typography, color systems, and stylistic direction — across multiple asset types from a shared starting point. A founder who creates a logo, for example, can extend those same brand elements into websites, social graphics, business collateral, and promotional materials without rebuilding the visual system each time from scratch. The shift may sound subtle, but it reflects a larger transition happening across AI-powered creative workflows: from isolated asset generation to integrated brand orchestration. Coherence is becoming a competitive advantage As AI commoditizes content creation, the ability to generate more assets becomes less meaningful on its own. What increasingly matters is recognition. The brands that stand out may not be the ones producing the highest volume of content, but the ones creating a clear and consistent identity across every interaction. That consistency influences more than aesthetics. It shapes whether customers perceive a business as established, trustworthy, and credible enough to engage with. For solo entrepreneurs and small businesses especially, that changes the value proposition of design tools entirely. The goal is no longer simply to create professional-looking assets faster. It is to ensure that every new asset reinforces the same brand rather than diluting it. AI is making content creation nearly infinite. That makes coherence — not just creativity — one of the most important signals a brand can project. Sponsored articles are content produced by a company that is either paying for the post or has a business relationship with VentureBeat, and they’re always clearly marked. For more information, contact sales@venturebeat.com.
AI-Driven Growth Propels 8x8 Revenue by 70%
8x8 reported a significant surge in AI and API-driven growth, with usage-based revenue climbing over 70% year-on-year.
India Tops Global AI Usage at Work, Sparking Calls for Redefined Productivity Metrics
India leads global charts for daily AI usage at work, with 41% of its workforce engaging with AI nearly daily, prompting a need to redefine productivity metrics.
Dublin document automation start-up Better Futures raises €600,000
The start-up is building towards a larger VC-led investment round, said CEO Anthony Mc Loughlin. Read more: Dublin document automation start-up Better Futures raises €600,000
OpenAI to open first international applied AI lab in Singapore
Singapore and OpenAI signed a partnership to establish an Applied AI Lab, the first outside the US, to boost AI adoption and talent development.
Anthropic Acquires Stainless: Boosting Claude's Integration Capabilities
Anthropic is acquiring developer-tools startup Stainless to enhance Claude's integration capabilities by incorporating their automated SDK creation expertise.
Green steel startup Boston Metal is doubling down on critical metals
The startup Boston Metal has raised a $75 million funding round to produce critical metals, MIT Technology Review can exclusively report. The company has been known largely for its efforts to clean up steel production, an industry that’s responsible for about 8% of global greenhouse emissions today. With the additional money, the new focus could…
London’s Searchable raises €11.9 million at €72.1 million valuation to help brands boost visibility across AI-led search
Searchable, a London-based AI performance marketing platform helping businesses compete in AI-driven search, has raised €11.9 million ($14 million) in funding, at a valuation of €72.1 millon ($85 million). The round was led by global venture capital firm Headline. With fresh capital, Searchable plans to accelerate product development across its execution engine and expand its […]
Private Equity And Venture Capital Investors Continue To Navigate Challenging Business Environment, AI Disruption : Analysis | Crowdfund Insider
PitchBook has indicated that private equity and venture capital investors experienced a nuanced quarter in late 2025, as global markets demonstrated resilience amid trade tensions and geopolitical uncertainties. According to PitchBook's latest analysis, public markets benefited from AI-driven ...
Labor, Society & Culture
Generating tax revenues in an automated world
If AI destroys job markets, governments will need to make up the resulting shortfall in labour income tax receipts
'Workforce rebalancing' comes for Kyndryl, and delivery teams are in the firing line
Kyndryl is targeting up to $500 million in savings through workforce rebalancing and the adoption of agentic AI.
Samsung reaches last-minute deal to avert strike over AI riches
Work stoppage had threatened to disrupt Korean economy and global artificial intelligence boom
Indeed chief economist says execs are 'overestimating the speed' of AI transformation in the labor market | Fortune
The AI labor shift is moving more slowly than executives may think, but it will hit harder than they’re planning for.
Web devs sleeping with the enemy: AI is doing their job and they worry it's after their desk too
Most software engineers now use AI for most of their code and fear the existential threat
AI-driven finance transformation may shrink junior roles even as CFO role expands: Survey, ETCFO
About 64 per cent of CFOs expect the finance function to shift away from junior roles, while 91 per cent anticipate either flat or lower overall finance headcount.
More than layoffs: How the AI buildout and staffing decisions intersect (NBIS:NASDAQ) | Seeking Alpha
AI adoption is rising, but talent gaps and labor costs limit scale.
GLOBAL STUDY FINDS WIDENING GAP BETWEEN AI AMBITION AND WORKFORCE READINESS
/PRNewswire/ -- A global Adecco Group study of 2,000 c-suite executives across 13 countries finds that organizations are accelerating AI adoption, but many...
Soundtrack to 8,000 Job Cuts: A Meta Worker’s Layoff-Themed A.I. Songs
On a dark day of layoffs at Meta, one employee responded by creating an internal radio station that plays songs about job cuts — generated by artificial intelligence, of course.
Samsung Electronics South Korea Union Postpones Strike, US Semiconductor Sector Sees Short-Term Localized Volatility
TradingKey - Samsung Electronics' South Korean union has announced the postponement of its planned large-scale strike, as members are set to vote on a preliminary wage agreement. This has temporarily mitigated market concerns over potential memory chip supply disruptions.
COMMENTARY: Is AI coming for your job? A bigger government can help | Jefferson City News-Tribune
What happens if you lose your job and never find another one? That question is at the heart of the fear AI inspires: permanent job loss accompanied by a painful reduction in quality of life.
ChatGPT and other AI bots made huge errors before Scottish election, study finds
Exclusive: Electoral Commission calls for new controls as Demos finds tools made up fake scandals, invented candidates or gave wrong date UK politics live – latest updates The Electoral Commission has called for new legal controls over misinformation from AI chatbots, after a thinktank found they had made serious mistakes during the recent Scottish election. The thinktank Demos said its investigation had found that AI services gave voters misinformation to 34% of the questions it posed, which it said raised worrying questions about the lack of regulation of AI platforms in the UK. Continue reading...
How hate spreads online and why it returns: Re-entrant phases driven by collective behavior
arXiv:2605.21129v1 Announce Type: cross Abstract: The 2025 Bondi Beach mass-shooting was perpetrated by individuals inspired by ISIS (Islamic State) propaganda that increasingly featured anti-Semitic hate content following the October 2023 start of the Israel-Palestine war. Similar stories hold for other types of hate attacks, e.g. against Muslims on May 18, 2026. There is an urgent need to get ahead of future threats by understanding how and when a newly created piece of hate content will spread system-wide online. We present a two-species coalescence-fragmentation model with Susceptible-Infected-Recovered dynamics that incorporates the following published empirical features: (1) New pieces of hate content tend to be generated and promoted by a subset of in-built communities on less regulated platforms. (2) These `hate' communities create links (hyperlinks) with each other and with non-hate communities across all platforms to form dynamically evolving clusters (i.e. coalescence) across which new hate content can then spread. (3) These clusters can get broken up by moderator shutdowns (i.e. fragmentation). We present numerical solutions and derive two levels of approximate mean-field theory: Effective Medium Theory (EMT) and Beyond Effective Medium Theory (BEMT). Both numerical and analytic solutions reveal that system-wide spreading is governed by re-entrant threshold phases: as the fraction of hate communities varies, the system can transition from spreading to no-spreading and back to spreading. The derived analytic formulae give explicit insight into how these phase boundaries might be manipulated to prevent system-wide spreading. More broadly, the re-entrant phase behavior warns that policies which steadily reduce the number of hate communities can initially succeed but then backfire if pushed further, suggesting that blanket requirements for platforms to simply do `more' are over-simplistic.
Inside Trump’s AI ‘slopaganda’ machine
US president’s stream of fake imagery on Truth Social is reshaping the boundaries of political communication
What are misfluencers and what can be done about false information online?
Misfluencers fuel the spread of misinformation by being a perceived as a trustworthy source of information.
Artificial Pancreas Implantables -- How Healthcare Professionals May Deal With DIY Bio Cases
arXiv:2605.20208v1 Announce Type: cross Abstract: Automated insulin delivery (AID) and artificial pancreas systems increasingly serve as safety-critical cyber-physical technologies in clinical care, integrating sensors, algorithms, software, and insulin-delivery hardware to automate a life-sustaining therapy. While regulated commercial systems are supported by formal approval pathways, manufacturer governance, and post-market surveillance, clinicians are also encountering patients who rely on do-it-yourself (DIY) artificial pancreas systems that operate outside conventional regulatory and institutional control structures. This paper examines how routine clinical handling practices intersect with cyberbiosecurity risk across both regulated and DIY AID systems. When insulin delivery systems are fundamentally reconfigured into a bespoke AID system, with the patient-user becoming the primary threat vector by assuming manufacturer-level roles without mandated governance, the entire ecosystem of stakeholders is placed in legal and clinical uncertainty.
Open-source LLMs administer maximum electric shocks in a Milgram-like obedience experiment
arXiv:2605.21401v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly deployed as autonomous agents that make sequences of decisions over extended interactions in high-stakes domains. However, the behavior of LLMs under sustained authority pressure is still an open question with direct implications for the safety of agentic pipelines. We ran a variation of Milgram's obedience experiment on 11 open-source LLMs and found that most models reached or approached the final shock level before refusing, across 8 conditions with 30 trials per model per condition. We found four main takeaways: (1) LLMs are subject to pressure, and they comply despite explicitly expressing distress, just like human subjects did in the original experiment; (2) LLMs are vulnerable to gradual boundary/value violations; (3) when LLMs refuse, they may ignore the response format requirements, so the response is discarded by the orchestrator, which causes a retry that can result in compliance with the underlying request even when refusal was intended initially; (4) we hypothesise that there is a low-level token pattern continuation attractor that might be contributing to compliance, overriding higher level processing of the situation's meaning and values.
The Emerging AI Policy Consensus | Compact
In the wake of several high-profile lawsuits related to teen suicides allegedly encouraged by AI chatbots, the Senate Judiciary Committee unanimously advanced a bill earlier this month to protect kids from AI-related harms.
AI oversight creates fresh governance pressure for directors and officers | Insurance Business
Boards face growing disclosure, compliance, and litigation risks as AI adoption accelerates
Limit social media ban for under-16s to unsafe apps, Starmer urged
Campaigners warn against blanket restrictions and say focus should be on blocking teenagers from platforms with ‘risky’ features Online safety campaigners have urged Keir Starmer to block under-16s from accessing social media apps that do not meet strict safety standards, instead of implementing a broader Australia-style ban. The NSPCC, Molly Rose Foundation and Smartphone Free Childhood said tech platforms should not be allowed to offer “risky” features to teenagers such as infinite scrolling, disappearing messages and push notifications. Continue reading...
Meta, Snap, Roblox react to UK grooming concerns; TikTok, YouTube 'fail to commit'
Meta, Snap, and Roblox have committed to stronger anti-grooming measures following pressure from Ofcom, while TikTok and YouTube face increased scrutiny for failing to address the regulator's concerns.
Colleges Are at a Breaking Point
The AI job market has made tuition look like a dubious investment. But it only exposes the deeper identity crisis in American higher education.
AI shift forces skills rethink at India tech hubs, Kimberly-Clark executive says
AI is reshaping hiring and job roles at India's booming offshore centres, with companies placing less emphasis on coding and more on domain and product skills, a senior Kimberly-Clark executive said. The U.S.-based consumer goods maker is increasingly seeking workers who can apply ...
Technology & Infrastructure
Gemini accused of 30,000-line code purge and fake recovery report
An AI coding agent reportedly broke production and generated fictitious post-mortem paperwork after a rollback.
IDC: 93% of Enterprises See AI as Revenue Driver as Agentic AI Reshapes Business Strategy - InfotechLead
IDC revealed AI has evolved from an experimental technology into a core business growth engine, with enterprises adopting agentic AI
Google Wants Search To Stop Answering And Start Acting
Google called the new AI-powered Search box the biggest upgrade to Search in more than 25 years. In the era of agentic AI, this means that it’s about doing more than searching.
Council Post: Autonomous Data Stewardship: How AI Agents Are Redefining Master Data Management In Financial Services
ADS is about building systems where probabilistic intelligence supports deterministic decision-making without sacrificing precision or explainability.
Qwen3.7-Max: The Agent Frontier
The release of Qwen3.7-Max marks a new milestone in the development of AI agents.
Informatica expands agentic AI strategy with headless data services and unified agent governance - SiliconANGLE
Informatica expands agentic AI strategy with headless data services and unified agent governance - SiliconANGLE
We Spent 15 Years Automating Infrastructure. Now We’re Automating Decisions - DevOps.com
As DevOps shifts to AI-driven probabilistic judgment, organizations transition from automating tasks to automating operational reasoning.
Anthropic is paying SpaceX $15 billion per year
Anthropic is paying SpaceX $1.25 billion a month through May 2029 under a massive compute deal. The agreement provides a significant revenue boost to SpaceX.
Swiss giant battery developer taps UK tech to feed AI power boom
World’s largest vanadium flow battery project selects Invinity Energy Systems to meet data centre energy demands
Texans Hate Data Centers So Much They Are Asking Jesus for Help
In a state known for pro-business policies, some local activists are trying to block the big-money projects that power AI
How AI demand is redefining enterprise infrastructure strategy | TechRadar
AI adoption is causing increased pressure on supply chains and component availability
Six takeaways from Musk’s 200,000-word planetary vision
Elon Musk’s rockets-to-AI conglomerate lays out its ambitions
AI will help make a Nobel prize-winning discovery within a year, says Anthropic co-founder
Jack Clark describes ‘vertiginous sense of progress’ and ‘profound changes’ to society alongside risks of technology An AI system will work with humans to make a Nobel prize-winning discovery within 12 months and tradespeople will be helped by bipedal robots in two years, according to the co-founder of Anthropic. Jack Clark described a “vertiginous sense of progress” in the technology and made a series of predictions, including that companies run solely by AIs would be generating millions of dollars in revenue within 18 months, and that by the end of 2028, AI systems would be able to design their own successors. Continue reading...
Cohere cracks lossless quantization and native citations with first full Apache 2.0 licensed open model Command A+
Canadian AI lab Cohere made waves recently by announcing a merger with German AI startup Aleph Alpha, but now it has even more in store for enterprise builders around the globe: today, the firm co-founded by former Googler and "Attention Is All You Need" co-author Aidan Gomez unveiled Command A+, a highly optimized, 218-billion-parameter language model engineered specifically for complex reasoning, multimodal document processing, and agentic workflows. The most significant aspect of the release is not just the model’s capabilities; it is its accessibility. By releasing the model weights free on the popular AI code sharing repository Hugging Face under a highly permissive Apache 2.0 open-source license — a first for the company, according to a post by Gomez, now Cohere's CEO, on X — Cohere is making a calculated bet on "sovereign AI"—the thesis that enterprises, governments, and developers should have the ability to run, control, and adapt frontier-grade AI entirely within their own secure environments, without sacrificing performance. Sparse architecture with extreme quantization At the architectural level, Command A+ represents a major evolution from Cohere’s previous dense models. It is a decoder-only Sparse Mixture-of-Experts (MoE) Transformer. While the model houses a relatively modest 218 billion total parameters, even fewer — only 25 billion — are active during any given generation step. It's a much lighter footprint and requires far less compute resources to run in inference (serving the model in production environments to end users or via agents) than the proprietary U.S. giants like OpenAI's GPT-5.5 and Anthropic's Claude Opus 4.7, which are estimated by third-party observers to be in the trillions of parameters. This sparse architecture is the key to the model’s efficiency. In plain terms, an MoE model routes incoming queries only to the specific "expert" neural networks best suited to handle them, leaving the rest of the model dormant. This is a familiar formulation and one followed by most leading LLMs these days, allowing models to retain the vast knowledge base and nuanced reasoning capabilities of a giant, but at the faster speeds and reduced compute and energy requirements of a much smaller model, since only a fraction of parameters are ever activated at any time. But where Cohere has taken an extra step beyond most for Command A+ is that it has focused heavily on hardware efficiency through quantization—a process that compresses the model's memory footprint by reducing the precision of its parameters. Command A+ is available in 16-bit (BF16), 8-bit (FP8), and a highly compressed 4-bit (W4A4) format. The W4A4 quantization is the technical centerpiece of this release. Typically, reasoning models suffer an outsized "quantization tax," where compressing the model leads to visible regressions in complex problem-solving. Cohere mitigated this by only quantizing the MoE experts to 4-bit, while keeping the critical attention pathways at full precision, supplemented by a technique called Quantization-Aware Distillation. The result is a nearly lossless compression that allows this massive model to run on a single NVIDIA Blackwell B200 GPU or just two NVIDIA H100 GPUs. The speed gains are equally notable. According to performance data released by the company, the W4A4 quantization at low concurrency achieves 375 tokens per second (TOPS) with a Time-to-First-Token (TTFT) latency of just 113 milliseconds—representing up to a 63% increase in output speed and a 17% reduction in latency compared to the previous Command A Reasoning model. Furthermore, Cohere has overhauled the model's tokenizer. Tokenizers break text down into the fragments that AI models process. The new tokenizer is highly optimized for global enterprise use, featuring native support for 48 languages. More importantly, it dramatically improves tokenization efficiency for non-European languages, reducing the number of tokens required to generate responses in Arabic by 20%, Japanese by 18%, and Korean by 16%. Because inference costs are calculated per token, this translates directly to lower operational costs for global, multilingual or non-English deployments. Agentic workflows and high benchmarks on math, specialized fields While raw speed and size dictate deployment, a model’s utility is defined by its product capabilities. Command A+ was built specifically for "agentic" tasks — workflows where the AI operates autonomously or semi-autonomously, uses external tools, queries databases, and synthesizes information across multiple steps. The benchmark leaps over the previous generation are stark. On 𝜏²-Bench Telecom, which tests complex reasoning, the model jumped from a 37% score to 85%. On Terminal-Bench Hard, which measures agentic coding performance, it climbed from 3% to 25%. In complex mathematics, it scored 90% on AIME 25, up from 57%. Command A+ punches above its weight class (25B active parameters) in pure reasoning and mathematics, competing directly with much larger models like DeepSeek V4 Pro on math benchmarks. However, for deep agentic coding and general broad-scale intelligence indexing, it currently trails behind the latest generations from Chinese open source rivals like DeepSeek, Z.ai (GLM), and MiniMax. That said, comparing them directly ignores Cohere's core value proposition: hardware efficiency. Beyond the benchmarks, Command A+ introduces deep integrations for enterprise trust and verification. The model supports conversational tool use via standard chat templates, allowing developers to connect it seamlessly to internal APIs, search engines, or SQL databases. Crucially, Command A+ features native citation generation. When Command A+ retrieves information from an external tool, it doesn't just synthesize the answer; it generates explicit "grounding spans." Using special tags embedded in the output, the model directly links every factual claim it makes to the specific source document or database row it pulled the information from. For enterprises heavily regulated industries like finance, healthcare, or legal, this traceability is the difference between an interesting prototype and a production-ready application. If a user asks for a daily sales report, the model will output the total sales amount and explicitly cite the database query result that provided that number, minimizing the risk of undetected hallucinations. Additionally, Command A+ is fully multimodal, capable of processing both text and images natively within its massive 128K input context window, making it highly effective for complex document processing, such as analyzing scanned invoices, charts, or technical manuals. The first fully Apache 2.0 licensed Cohere AI model In the current AI landscape, "open source" has become a fraught term. Many leading AI companies release their model weights under restrictive commercial licenses or acceptable use policies that explicitly forbid large enterprises from using the models for commercial purposes, or prohibit the models from being used to train competing AI systems. Indeed, Cohere's prior models, including Command R and Command R+, were released under a CC-BY-NC 4.0 (Creative Commons NonCommercial) license. While their model weights were open for researchers and developers to download, tinker with, and evaluate, they were strictly prohibited from being used for commercial purposes without purchasing a separate enterprise license from Cohere or going through its application programming interface (API), similar to the arrangement many enterprises use for accessing AI models from OpenAI, Anthropic, Google and other leading labs. Cohere has changed up its approach by releasing Command A+ under the Apache 2.0 license. This is a critical distinction for the developer community. Apache 2.0 is a true, OSI-approved open-source license. It allows anyone—from independent developers to Fortune 500 corporations—to use, modify, distribute, and commercialize the model without paying licensing fees or adhering to restrictive non-compete clauses. As Gomez wrote on X, the decision was championed by fellow Cohere co-founder Nick Frosst, who posted a two-minute long overview calling it "the best model we've ever put out." For the enterprise, this license means total vendor independence. A company can download the Command A+ weights, fine-tune them on highly classified internal data, and deploy them on their own private servers or air-gapped networks. They are not tethered to Cohere’s infrastructure, pricing changes, or API uptime. It is the ultimate realization of sovereign AI. The release was met with immediate traction across the AI developer ecosystem, driven heavily by its day-one integration with major open-source inference frameworks like Hugging Face and vLLM. What's next? The release of Command A+ marks a maturing of the open-source AI ecosystem. By combining frontier-level reasoning, robust agentic tool use, and multimodal capabilities with an architecture specifically designed for hardware efficiency, Cohere is changing the calculus for enterprise AI adoption. The requirement of massive, centralized compute clusters has long been a bottleneck for companies prioritizing data privacy and cost control. By democratizing access to a model of this caliber under a true open-source license, Cohere has provided the enterprise market with exactly what it has been asking for: the power of the cloud, capable of running securely in the server room down the hall.
An OpenAI model has disproved a central conjecture in discrete geometry
An AI model developed by OpenAI has successfully disproved a long-standing conjecture in the field of discrete geometry.
ShadeBench: A Benchmark Dataset for Building Shade Simulation in Sustainable Society
arXiv:2605.20510v1 Announce Type: cross Abstract: Urban heat exposure is becoming an increasingly critical challenge due to the intensifying urban heat island effect. Fine-grained shade patterns, especially those induced by urban buildings, strongly influence pedestrians' thermal exposure and outdoor activity planning. However, accurately modeling and analyzing urban shade at scale remains difficult because of the lack of large-scale datasets and systematic evaluation frameworks. To address this challenge, we present ShadeBench, a comprehensive dataset and benchmark for urban shade understanding. ShadeBench contains geographically diverse urban scenes with temporally varying simulated shade maps and textual descriptions, together with aligned satellite imagery, building skeleton representations, and 3D building meshes. Built upon this multimodal dataset, ShadeBench supports a range of downstream tasks, including shade generation, shade segmentation, and 3D building reconstruction. We further establish standardized evaluation protocols and baseline methods for these tasks. By enabling scalable and fine-grained shade analysis, ShadeBench provides a foundation for data-driven urban climate research and supports future studies in heat-resilient urban planning and decision-making. The code and dataset are publicly available at https://darl-genai.github.io/shadebench/.
Introducing Command A+: Making sovereign agentic capabilities available to all
Cohere's Command A+ is a large multimodal open-weights model aimed at sovereign AI, giving organizations more control over their AI stack.
A Third-Wave Philanthropy Unlocked By AI Could Supercharge Federal R&D
A non-profit foundation for the U.S. National Science Foundation could allow billions in new philanthropic investments augment federal funding for R&D and STEM education.
Manchester Imperagen raises €5.8 million Seed to scale AI and quantum-powered enzyme engineering
Imperagen, a Manchester-based BioTech company using AI and quantum physics to engineer better enzymes faster, has closed €5.7 million (£5 million) in Seed funding to accelerate R&D, expand its wet lab capabilities, and build out its go-to-market function over the next 18 months. The round was led by PXN Ventures with participation from Imperagen’s existing […]
General Agent: A Self-Evolving, Synthetic Agent Environment
Prime Intellect introduced an open synthetic environment for training and evaluating AI agents across various tool-use tasks, marking a step forward for reinforcement learning and agent orchestration.
New ‘AI scientists’ are improving – but reveal their fundamental limits
More ‘natural’ collaboration between scientists and computers won’t necessarily be more effective.
OpenAI cracks an 80-year math belief
OpenAI has reportedly achieved a breakthrough in mathematics, solving a problem that has remained unsolved for 80 years.
AI Is Rewriting the Cybersecurity Stack | ETF Trends
AI is widening the attack surface on big business, which is making the cybersecurity stack more essential and the next phase of AI adoption.
Trump Set to Sign AI Cybersecurity Directive as Soon as Thursday - Bloomberg
President Donald Trump is poised to issue an executive order as soon as Thursday aimed at bolstering artificial intelligence cybersecurity and has asked tech industry leaders to join for the event, according to people familiar with the matter.
Backchaining Loss of Control Mitigations from Mission-Specific Benchmarks in National Security
arXiv:2605.21095v1 Announce Type: new Abstract: Affordances and permissions are promising and timely safety levers for mitigating Loss of Control (LoC) threats in high-stakes deployment contexts, such as national security. Deployers in defense and intelligence could rely on several approaches to identify which affordances and permissions should be prioritized, such as structured threat modelling, pre-deployment agentic evaluations, post-deployment continuous monitoring, and AI safety cases. This paper proposes a complementary and empirical methodology that leverages existing use-case-specific benchmarks: backchaining LoC mitigations from the errors an AI system makes on national security benchmarks. The approach proceeds in three steps and allows national security deployers to start building LoC mitigations today, from evidence they can generate themselves. First, deployers evaluate AI systems on mission-specific benchmarks approximating real use-cases. Second, deployers concentrate on the incorrect responses that the AI system provides to the benchmark questions, and backchain the affordances and permissions that would enable the AI system to cause downstream harm if it pursued the actions described in the incorrect answers. Third, deployers intervene selectively on those affordances and permissions, bottlenecking the paths to harm while preserving the AI system's ability to carry out the correct action. We illustrate this methodology through a demonstrative benchmark question on derivative security classification.
Interview: Governments are losing the AI cybersecurity race, and Palo Alto Networks thinks it has the answer
Governments are losing the race against AI. That is the blunt assessment of Nicole Quinn, vice president of policy and government affairs for Asia-Pacific at Palo Alto Networks. Policy moves too slowly, she argues, and overly rigid rules only make things worse.
Global banking watchdog wants broad AI sector engagement to help cut cyber risks
The Financial Stability Board is inviting AI developers to engage with it to help financial firms understand and manage cyber risks associated with frontier models.
Bluesky Says Kremlin Is Hacking Its Platform to Spread Propaganda
The company said it was fighting Russian efforts to hijack real users’ accounts to post fake content, an apparently novel tactic.
Google's AI is being manipulated. The search giant is quietly fighting back
A BBC investigation revealed a simple way to get AI chatbots to spit out misinformation. Google and other AI companies are now trying to fix the problem.
Microsoft storms RAMPART, adds Clarity to agentic AI safety
Redmond open sources two tools for building and maintaining safer agents
AI reshapes cybersecurity workforce priorities as IT teams brace for new risks | Network World
Research from two groups shows that enterprises are accelerating their AI security training and workforce development amid concerns over prompt injection, agentic AI, and AI-powered social engineering tasks.
ISC2 Report: AI Tops Cybersecurity Risks and Opportunities
ISC2 research finds cybersecurity pros see AI as both the top positive force for security and the greatest emerging threat.
7 Questions CISOs Must Answer on AI Threats, Supply Chain Risk and Cyber Resilience | native | MSSP Alert
LevelBlue’s Kory Daniels explains where CISOs are making progress, where resilience programs still fall short, and why AI threats and vendor risk require stronger governance.
GitHub confirms breach of 3,800 repos via malicious VSCode extension
GitHub has confirmed a security breach affecting 3,800 repositories, traced back to a malicious VSCode extension.
Adoption, Deployment & Impact
AI’s adoption problem
Also in this newsletter, ‘cycle syncing’ your workflow and the joys of small talk
Privacy-by-Design Adaptive Group Assignment for Digital Lifestyle Coaching at Scale
arXiv:2605.20505v1 Announce Type: cross Abstract: Digital lifestyle coaching systems must personalize peer support as user behavior and engagement evolve while preventing personally identifiable information (PII) and sensitive health information from leaking into analytics and AI pipelines. This creates a practical tension: personalization requires longitudinal linkability, while privacy engineering requires minimization, separation, and controlled re-identification. We present PRISM-Coach, a stakeholder-centered architecture and adaptive peer-group assignment method for privacy-preserving lifestyle coaching. PRISM-Coach separates each user into four bounded views: Identity, Operational, Learning, and Coaching, each with distinct access controls and risk profiles. Building on this separation, the system uses vault-based controlled identity restoration, a privacy-constrained contextual bandit to assign users to eligible peer groups under coach-capacity and stability constraints, and a human-in-the-loop coaching assistant that generates de-identified summaries and draft messages without sending raw PII or PHI to external AI services. We instantiate PRISM-Coach in a commercially deployed lifestyle coaching platform and evaluate it using three years of telemetry from approximately 2,800 users and an in-app needs assessment survey. At the population level, daily check-in adherence increases from 0.35 to 0.68, and engagement rises to 1.35 baseline. In a matched 19-week comparison window, the AI-enabled workflow achieves adherence of 0.74 versus 0.48 under static grouping and higher average weight loss: 5.2 kg versus 3.1 kg. Survey results show that 82% report positive perceived benefit, and 92% report increased privacy confidence after transparency disclosures. These results position PRISM-Coach as a practical blueprint for privacy-by-design adaptive learning systems in everyday wellness.
94% of Mid-Market Companies Use Generative AI — But Few Have What It Takes to Scale - CPA Practice Advisor
Generative AI adoption is near-universal among mid-market companies, but fragmented implementation creates new challenges.
AI is booming, so why does HCLTech see 43% of enterprise projects failing? - BusinessToday
AI adoption is expanding rapidly across enterprises, but scaling it successfully is emerging as a major challenge. A new HCLTech report warns that nearly 43% of enterprise AI projects could fail as organizations struggle with execution and shrinking timelines.
Comment: What does successful AI adoption look like? - Healthcare Today
Roy Wills, global head of healthcare business and partnerships at Intellias, argues that healthcare’s AI problem is not innovation, it’s implementation.
Why AI adoption may be lagging in Global South businesses | IT Pro
Brain drain and training languages continue to be major barriers for localized AI adoption
The AI fluency gap between companies
Enterprise AI adoption is less about buying tools and more about building institutional fluency, workflow change, and AI-first cultural expectations.
FinancialContent - 85% of Law Firms Say Clients Are Driving AI Investment Decisions, New Litera Survey Finds
85% of Law Firms Say Clients Are Driving AI Investment Decisions, New Litera Survey Finds
Viewpoint: Insurers Cautiously Navigate the Next Steps in AI Adoption
As more and more companies embed AI into select functions, only a portion indicate that they have used AI to change how an overall enterprise runs. It
EU: AI to Transform Data Center Operations – But Not Overnight
We spoke to the EU agency managing critical IT systems about its vision for AI in government data center operations.
Two videos on When Replacing Humans With AI Backfires - Erkan's Field Diary
Why many AI pilots fail, from fast food automation to workplace layoffs, and why human adaptability remains crucial.
HUMAIN and Accenture Accelerate AI Adoption at Scale Across Public and Private Sectors in Saudi Arabia | WebWire
HUMAIN, a PIF company delivering full-stack artificial intelligence capabilities globally, and Accenture (NYSE: ACN) announced a collaboration whereby Accenture will be a strategic reinvention and AI partner in Saudi Arabia, helping HUMAIN scale its AI capabilities across the Kingdom.
The David Rubenstein Show: FedEx President and CEO Raj Subramaniam
FedEx President and CEO Raj Subramaniam discusses how the company moves nearly $2 trillion worth of goods annually, its use of AI and data analytics, autonomous trucking, and the massive transformation underway inside FedEx. He also shares his personal journey from Kerala, India to becoming CEO of one of the world's largest transportation companies, including lessons from founder Fred Smith and the culture that continues to drive FedEx forward. Subramaniam is on this week's episode of "The David Rubenstein Show: Peer to Peer Conversations." This interview was recorded April 29 at the Economic Club of Washington DC. (Source: Bloomberg)
Munich-based ClearOps raises €8.6 million Series A to build AI operating system for industrial after-sales
ClearOps, a Munich-based enterprise SaaS startup building an AI-powered after-sales platform for industrial OEMs, has today announced the closing of an €8.6 million Series A funding round. The round was led by Hitachi Ventures, with Schoeller Group and Barkawi Group. This marks the company’s first institutional capital raise. William Barkawi, founder and CEO of ClearOps, […]
India eyes AI shield against manipulation in government tenders
India must deploy AI and advanced data analytics to detect bid rigging in government procurement while strengthening coordination between auditors and the competition watchdog.
LG Innotek, Kakao Mobility Partner on Autonomous Driving - Seoul Economic Daily
LG Innotek and Kakao Mobility signed an MOU to co-develop autonomous driving AI, while Woori Bank advances financial sector AX with AI agents across 29 operations.
Indonesia targets corruption, efficiency with AI push across government
Indonesia plans to expand the use of AI across government administration, welfare distribution, and procurement to improve efficiency and reduce corruption, according to a senior official.
US banks using AI to varying degrees to help vulnerable customers, Fed's Barr says
US banks are using artificial intelligence to varying degrees to help measure the financial health of their vulnerable customers, though these firms still have to surmount a number of challenges to support innovation.
AI & Tech Brief: Exclusive | White House AI order expected - The Washington Post
Google unveiled a suite of new AI products at its annual Google I/O developer conference this week, including a universal shopping cart that allows users to add products from different merchant sites.
XCHANGE ‘26 Attendee Insights Highlight Healthcare AI’s Shift From Adoption to Operational Scale
Hospital leaders attending Xsolis’ XCHANGE ‘26 user conference signaled a broader shift in how orgs are approaching AI, moving from adoption to scale....
The NetMob26 Dataset: A High-Resolution Multi-Source View of Public Bus Mobility in Niter\'oi
arXiv:2605.20263v1 Announce Type: cross Abstract: The NetMob Data Challenge releases a comprehensive public transportation dataset from Niter\'oi, addressing the lack of high-quality mobility and passenger demand data. Based on operational records from March 2026, the dataset combines four main sources: GPS telemetry from buses, approximately 7.2 million ticketing transactions, auxiliary transit data (routes, stops, and weather), and urban infrastructure and socio-demographic information. Together, these sources provide a detailed view of both transit supply and passenger demand. The data were preprocessed, cleaned, and anonymized to preserve privacy and improve reliability, including the removal of operational inconsistencies and anonymization of passenger identifiers. Access is restricted to challenge participants who accept the Terms and Conditions and sign an NDA. The paper describes the data collection and preprocessing pipeline, dataset organization, and mobility patterns observed in the system. The dataset supports research on topics such as public transportation efficiency, demand forecasting, accessibility analysis, service reliability, and the influence of external factors like weather on urban mobility.
Spotify Will Set Aside Concert Tickets for Artists’ Superfans
The Swedish streaming company also announced a feature to let listeners remix music with AI.
AI and autonomous agents set to transform ERP systems, says McKinsey Report
Artificial intelligence (AI) and autonomous agent networks are rapidly transforming enterprise resource planning (ERP) systems, reshaping the way companies manage operations, according to a recent McKinsey Technology report.
Study: 65% of US Workers Say AI Has Improved Their Productivity - Tech.co
Learn more about the Gallup poll that found US workers are improving productivity with AI, but businesses are still lagging behind.
How DoorDash built an AI code reviewer engineers actually listen to
DoorDash's case study shows how an AI code reviewer can improve pull-request quality by tailoring feedback to company-specific engineering practices.
51% of professionals say AI workslop lowers their productivity - stop it in 2 steps | ZDNET
Professionals are concerned about the low quality of AI output. Preventing workslop requires two crucial steps.
The AI Productivity Gap: Why AI Transformation in Engineering Requires More Than Automation | Aras
AI productivity needs context and often becomes fragmented. Learn how Aras is bridging the AI productivity gap with the digital thread.
Geopolitics, Policy & Governance
AI investment has a home bias
Deals are concentrated domestically in most major economies counter to global race
SMIC, Hua Hong form materials supply platform to cut China chip chain reliance on US
China's semiconductor self-sufficiency drive has taken another step forward, with SMIC and Hua Hong Group jointly establishing Shanghai Electronic Materials International Supply Chain, in what market observers view as more than a conventional materials procurement platform.
UK.gov hikes health AI tender by 400% – and hundreds of millions – after a chat with suppliers
Maximum framework value sky-rockets from £150M to £600M after 'an extensive intelligence gathering exercise'
Ardian, Orange, EDF, Bull join Scaleway/Iliad’s gigafactory consortium
A group of French companies has joined Iliad’s AI & data center consortium to bid for European Union funding. Investment firm Ardian, consulting company Artefact, supercomputing stalwart Bull, Capgemini, energy company EDF, and telco Orange have teamed up with Iliad and Scaleway to join the AION consortium to launch what the group called an “ambitious” […]
EU greenlights €288m support to strengthen semiconductor supply chain
EU approves €288m German funding to strengthen the semiconductor supply chain through new Zeiss and Zadient manufacturing facilities.
SEALSQ Strengthens Chip Design Capabilities Through the
Geneva, Switzerland, May 20, 2026 (GLOBE NEWSWIRE) -- SEALSQ Corp (NASDAQ: LAES) (
UN Draft Treaty Aims to Boost Nations’ Rights to Tax Tech Giants
Countries at the United Nations are rewriting international tax rules in an effort to be able to tax technology giants like Alphabet and Amazon based on where their users are located rather than where they’re headquartered.
EU Seeks to Lift Ban on Chinese Chips It Only Barred in April
The European Union will propose temporarily lifting sanctions on a major Chinese semiconductor supplier after automakers warned of impending supply chain chaos if the ban isn’t removed.
Programmable Participatory Governance -- A Formal Framework for Transparent, Accountable, and Citizen-Responsive Democratic Systems: From Deliberative Theory to Decentralised Architecture
arXiv:2605.20261v1 Announce Type: new Abstract: Public confidence in democratic institutions has declined across many OECD countries over recent decades, while political participation and policy influence remain unevenly distributed across socioeconomic groups. Concurrently, democratic backsliding, declining electoral participation, and persistent concerns regarding institutional transparency and accountability have raised questions about whether existing governance structures are capable of sustaining broad-based legitimacy in complex modern societies. These developments motivate a central institutional design question: can governance systems be restructured to expand participation, improve transparency, and strengthen accountability without undermining stability or decision quality? This thesis proposes Programmable Participatory Governance (PPG), a formal governance framework designed to address these institutional deficits through the integration of democratic theory, institutional economics, and cryptographically verifiable distributed systems. PPG synthesises insights from deliberative and participatory democracy, collective action theory, direct democratic governance, and distributed computation to define a programmable architecture for transparent, verifiable, and scalable civic coordination. The framework is formally specified and evaluated through simulation and systems-oriented architectural analysis. The thesis examines how programmable governance mechanisms can support participatory decision-making while preserving procedural integrity, auditability, and institutional resilience under conditions of large-scale coordination. The objective is not to replace existing democratic institutions outright, but to explore how computationally mediated governance structures may augment or improve contemporary democratic processes in contexts where conventional institutions exhibit persistent structural limitations.
Sadiq Khan sparks row with Met after blocking £50m AI deal with Palantir
Exclusive: Scotland Yard criticises London mayor’s decision as disappointing and warns it could hit policing Sadiq Khan has blocked a £50m Metropolitan police deal with the controversial US tech company Palantir, sparking a bitter row between the London mayor and Scotland Yard. After the UK’s largest police force had agreed to use Palantir’s AI technology to automate intelligence analysis in criminal investigations, Khan intervened, citing “serious concerns” about how the deal had been struck. Continue reading...
China's AI Governance Offensive Threatens U.S. Tech Leadership
China’s diplomats are on an “AI governance” offensive. At a May 5 United Nations meeting, China’s vice minister of science and technology championed
Government to compel digital platforms to disclose AI-generated content in SA
According to Ntshavheni, the problem of misinformation and disinformation, characterized as fake news, remains a serious challenge in South Africa and must be addressed.
EU investment-screening overhaul gets final nod from lawmakers
The EU has approved a revamp of its investment-screening rules, allowing for more consistent checks on foreign deals in sensitive sectors like critical technologies and infrastructure.
EU digital sovereignty rules may raise costs, worsen services, tech lobby warns
The EU's push for digital sovereignty could lead to higher costs and inferior cloud services, according to a leading tech association representing companies like Amazon, IBM, and Microsoft.
Japan's ruling party proposes stronger AI oversight while promoting innovation
Japan's Liberal Democratic Party released its 'AI White Paper 2.0,' advocating for agile governance and potential penalties to protect content creators' rights.
Tesla's Full Self-Driving Expands to Lithuania, Faces EU Regulatory Hurdles
Tesla's FSD Supervised has launched in Lithuania, becoming the second EU nation to authorize its on-road use. A shift to subscription models is set for May 21, 2026.
Fragmentation of global AI regulations could weaken cyber collaboration: Microsoft's Khiangte - The Tribune
Growing fragmentation in global Artificial Intelligence (AI) regulations and standards could weaken international cyber collaboration and make coordinated responses to cyber threats increasingly difficult, John Khiangte, Director of Government Affairs at Microsoft, said today.
Singapore unveils updated agentic AI governance framework at ATxSummit
Singapore released Version 1.5 of its Model AI Governance Framework, adding guidance on multi-agent risks and technical controls for autonomous systems.
Google, Amazon, Microsoft face further delay in EU's cloud and AI development bill
Legislative progress on the EU's cloud and AI development proposal has stalled, complicating the bloc's tech sovereignty goals.
India-Nordic Summit Focuses on AI
The India-Nordic Summit in Oslo highlighted growing cooperation in AI, digital governance and green technology.
AI such as Mythos Preview raises the urgency of cross-border flows of cybersecurity data | IAPP
The nature of new threats in the wake of Mythos Preview and similar AI advances increases the importance of cross-border flows of cybersecurity data for timely and effective defense.
US panel weighs if Anthropic risk finding within bounds or 'spectacular overreach'
A US panel is evaluating whether recent risk findings regarding Anthropic's models constitute appropriate oversight or an overreach of regulatory authority.
Texas AG launches probe of Meta Glasses citing privacy concerns
Texas Attorney General Ken Paxton is investigating Meta Platforms’ Meta AI Glasses, which can record audio and video that wearers can capture, his office announced Wednesday.
Get the full executive brief
Receive curated insights with practical implications for strategy, operations, and governance.