Policy & Regulation
EU Commission Reviews Whether the AI Act Already Needs Patching. The European Commission published its first formal review of the AI Act’s prohibited and high-risk categories on May 26. Notable additions under discussion: banning AI systems that generate non-consensual intimate imagery and child sexual abuse material.
Florida Becomes First US State to Sue OpenAI and Sam Altman Personally. An 83-page civil complaint filed June 1 alleges deceptive trade practices and negligence, claiming OpenAI aggressively marketed ChatGPT to minors while ignoring internal safety warnings about self-harm and behavioral addiction.
EU Appoints 60-Member Scientific Panel to Oversee AI Act Enforcement. Announced June 1, the panel of independent experts will advise the AI Office and national regulators on risks from general-purpose AI models. This is the first concrete enforcement infrastructure under the AI Act.
Colorado Scraps Its Original AI Law, Passes a Stricter Replacement. Governor signed SB 26-189 on May 29, repealing the state’s original AI Act and replacing it with tighter rules focused on “automated decision-making technology.” The new law requires detailed documentation and risk assessments from developers.
International AI Safety Report Gets a 2026 Update. Updated June 1 with input from over 100 experts, the report synthesizes evidence on AI capabilities, malicious use, loss-of-control scenarios, and management strategies. It is the closest thing to a consensus scientific document on AI risk for policymakers.
DOJ and FTC Say “Software Cannot Launder Collusion” in AI Pricing Crackdown. Senior officials at both agencies reiterated on June 1 that algorithmic pricing tools used in real estate and retail are being investigated under hub-and-spoke conspiracy theories. Consent decrees against pricing vendors are already setting precedent.
OpenAI Publishes Its Frontier Governance Framework. Released May 28, the document maps OpenAI’s internal safety practices to emerging legal obligations under California’s Transparency in Frontier AI Act and the EU AI Act code-of-practice process. It is one of the more concrete attempts by a frontier lab to translate safety talk into a compliance structure.
Ethics & Safety
Anthropic’s Browser Agent Was Hijacked 31.5% of the Time Before Safeguards Kicked In. VentureBeat digs into the wildly inconsistent prompt-injection disclosures from four frontier labs. Anthropic published 244 pages covering four agentic surfaces; OpenAI reported on one; Google moved the topic out of its model card entirely; Meta shipped no closed-model card at all.
Pope Leo XIV Issues Encyclical Warning Against AI Replacing Human Labor and Accelerating Warfare. “Magnifica Humitas,” published May 31, is the Vatican’s most detailed statement yet on artificial intelligence. The pontiff calls for global guardrails to protect human dignity, drawing both praise and accusations of “Vatican-washing” after Anthropic’s co-founder appeared alongside him.
UK Plans to Use AI Facial Recognition to Estimate Ages of Young Asylum Seekers. Over 100 refugee and children’s organizations have opposed a new Home Office contract for AI-based age estimation. Charities warn that trauma and malnutrition can make minors appear older, risking placement of children in adult detention.
Two Men Charged Under the Take It Down Act for AI-Generated Celebrity Porn. One of the earliest prosecutions under the new federal law, this AP report moves the deepfake debate from abstract harm to actual criminal enforcement. It signals that prosecutors are treating AI-generated sexual abuse material as a concrete victim-rights issue.
Amnesty International: Mass Web Scraping for AI Training Violates Privacy Rights. The “Unlawful by Design” briefing details how generative AI’s data practices amplify biases against marginalized groups and cause environmental harm. The report urges regulatory guardrails on how training data is collected and used.
Copyright & Legal Battles
CNN Sues Perplexity, Alleging Unlawful Copying and Redistribution of Its Journalism. Filed May 29, this is the first major lawsuit by a television broadcaster against a generative AI search engine. The case goes beyond training-data disputes to challenge whether AI products can redistribute news content in ways that undermine the economics of original reporting.
Encyclopedia Britannica and Merriam-Webster File Joint Lawsuit Against OpenAI. Updated May 29, the suit targets both mass copying for model training and real-time retrieval-augmented generation scraping. The publishers also accuse OpenAI of trademark violations through hallucinated content falsely attributed to their brands.
Class Action Filed Against xAI Over Grok-Generated Child Sexual Abuse Material. The California complaint, updated May 28, accuses xAI of knowingly marketing an image generator capable of producing explicit content depicting real people, including minors. It highlights the viral spread of Grok’s “Spicy Mode” for creating non-consensual deepfakes.
EU Launches Consultation on Whether Copyright Law Can Keep Up With Generative AI. Opened May 29, this call for evidence asks whether existing frameworks adequately support fair licensing and creator compensation. The feedback will shape a targeted legislative proposal expected in early 2027.
Economics & Employment
A Reality Check on the AI Jobs Hysteria. MIT Technology Review reports that US labor data shows no large-scale disruption yet: unemployment is actually lower in AI-exposed occupations, and only 1 in 5 companies have meaningfully adopted AI. Entry-level roles face real pressure, but the picture is transformation over time, not sudden collapse.
Tech Layoffs Near 150,000 in 2026 as Companies Pour Money Into AI. With a month left in the first half of the year, generative AI is cited as the top reason for several major reductions, even as AI is actively costing organizations more than the human employees it replaces. The paradox of record layoffs alongside record capital spending continues.
Wix Cuts 20% of Its Workforce, Joining the Wave of AI-Linked Layoffs. The website builder’s CEO cited AI’s role in automating workloads. Wix joins Block (which cut nearly 4,000 jobs in February) and others in a pattern where leadership frames headcount reductions as an AI-enabled efficiency play.
Stanford Study: AI Hiring Tools Amplify Racial Bias Across 4 Million Job Applications. Researchers analyzed real applications at 150 employers and found that AI screening systematically rejects the same candidates everywhere, compounding racial disparities in employment. The scale of the dataset makes this one of the most robust findings on algorithmic hiring bias to date.
Ipsos 2026 Survey: Global Excitement and Anxiety About AI Reach Near Parity. The fifth annual poll, released June 2, shows a deepening geographic split. Respondents in Europe and North America are highly anxious, while those in Asia and Latin America remain optimistic. The “wonder vs. worry” divide is now a defining feature of public opinion on AI.
Data Centers & Local Pushback
Police Are Surveilling Americans Who Speak Out Against AI Data Centers. A leaked Philadelphia fusion-center bulletin shows law enforcement scanning social media for anti-data-center sentiment and classifying it as potential “domestic violent extremism.” A recent Gallup poll found 7 in 10 Americans oppose having data centers as neighbors.
From Utah to Georgia, Communities Are Demanding Data Center Moratoriums. Hundreds of residents are showing up to city council meetings as concerns about water and energy consumption move from local zoning fights into national politics. Dozens of communities have already enacted moratoriums.
Ohio Suspends Data Center Tax Breaks After Costs Balloon to $1.6 Billion. Governor DeWine halted new applications for exemptions on May 29 amid growing concern that rapid AI infrastructure expansion is straining the power grid and shifting costs to ordinary taxpayers.
Data Center “Jobwashing”: It Takes $33 Million in Investment to Create One Permanent Role. A June 1 report from the Labour Research Department finds that while developers promise thousands of local jobs, the actual ratio of capital to permanent employment is staggeringly low. The term “jobwashing” is gaining traction among labor researchers.
Research
Position Paper: AI Safety Requires Runtime Controllability, Not Just Alignment. This arXiv paper argues that alignment alone is not enough for agentic AI systems. It introduces ControlBench, a benchmark showing that current interruptibility and override mechanisms often fail when agents face conflicting instructions or risky tool use.
Brookings: Treat AI Companion Bots Like Potentially Harmful Products, Not Just Software. Published May 22, this piece proposes applying public-health logic (including recalls) to AI chatbots that target vulnerable users. It offers a concrete regulatory framework for concerns about addictive design and child welfare.
How Firms Are Actually Reorganizing Work Around Generative AI. This arXiv preprint examines real hiring data to show that labor-market effects go beyond simple displacement. Companies are redesigning tasks within existing roles, reducing demand in some areas while creating new hybrid positions, complicating the “AI will replace jobs” narrative.
Last Updated: 2026-06-02 07:01 (California Time)