The Data Center Backlash
Seven in Ten Americans Oppose Local AI Data Centers, Gallup Finds. A Gallup poll shows broad public opposition to data center construction, with nearly half of respondents “strongly” opposed. Resource consumption, especially water and energy use, is the top concern. Rural communities in Utah and elsewhere are organizing referendums to overturn local approvals.
Anti-Data Center Activists Are Taking on Big Tech and Winning. More than 70 data center projects have been rejected or restricted in the first four months of 2026 alone, exceeding the total for all of 2025. Compass Datacenters withdrew plans for an 800-acre project in Prince William County, Virginia after intense local pushback.
Women Are Leading the Fight Against AI Data Centers. Women are disproportionately driving opposition to data center construction across the country, according to recent polling. Environmental justice advocates say Big Tech is targeting Black, Brown, and Indigenous communities already overburdened by industrial pollution.
Opposition Groups Have Blocked or Delayed $64 Billion in Data Center Projects. Data Center Watch estimates at least 142 activist groups are now operating across 24 states. In Warrenton, Virginia, residents voted out every town council member who supported a proposed Amazon data center campus.
Policy & Regulation
Illinois Passes First-in-Nation Bill Requiring Third-Party Safety Audits of Frontier AI. SB 315 passed the Illinois legislature and would require developers of large AI models to publish annual risk plans, undergo independent safety audits, and implement whistleblower protections. OpenAI and Anthropic publicly supported the bill, which now heads to Governor Pritzker’s desk.
Why Trump Pulled His AI Safety Executive Order at the Last Minute. The White House shelved a planned executive order that would have created a voluntary pre-deployment safety testing framework for frontier AI models. Industry leaders reportedly called the president directly, and concerns about competitiveness with China won out over safety advocates.
Newsom Signs Executive Order Framing AI as a Labor Disruption Issue. California’s new order focuses on worker retraining, higher-education alignment, and links labor policy to adjacent harms like deepfakes and digital likeness protection. It treats AI primarily as a workforce challenge rather than a pure technology policy question.
EU Reaches Deal on AI Omnibus, Bans Nudification Apps. EU negotiators agreed on the first round of amendments to the AI Act, postponing some compliance deadlines for high-risk systems to December 2027. The deal adds new prohibitions on AI-generated non-consensual intimate imagery and child sexual abuse material.
Missouri Bans AI Therapy Chatbots, Vermont Recognizes Neurological Rights. Missouri lawmakers passed a bill prohibiting AI chatbots from advertising themselves as capable of offering therapy or mental health diagnoses, with $10,000 fines for first offenses. Vermont became the first state to formally recognize personal neurological rights.
State AI Legislation Tracker Shows Massive Activity Across the Country. This weekly roundup documents a wave of state-level AI bills. California alone passed nine bills through a chamber in one week, covering AI transparency, employment, and healthcare. Illinois advanced consumer data privacy, chatbot safety, and algorithmic pricing transparency measures.
Canada to Release National AI Strategy Focused on Democracy and Sovereignty. Prime Minister Carney announced the long-delayed strategy will cover six pillars including protecting democracy, building sovereign compute infrastructure, and addressing labor market disruptions. The announcement comes amid growing public concern over AI’s societal effects.
Ethics & Safety
First Federal Charges Filed Under AI Revenge Porn Law. Two men face up to two years in prison under the 2025 Take It Down Act for using AI to generate sexually explicit images of women without consent. Court filings allege one defendant published at least 360 albums containing AI-generated pornography of around 90 different women.
How AI Deepfakes Tore a Pennsylvania High School Apart. This investigative report documents what happens when AI-generated sexual abuse material hits a real school community, exposing failures across institutions from police to administrators. It moves the deepfake conversation from abstract risk to concrete, local harm.
The Blind Spot in AI Safety: When Models Drift From Reality Without Anyone Noticing. This analysis critiques a recent Anthropic paper arguing that AI failures will be unpredictable accidents rather than coherent misalignment. The author warns that “epistemic drift,” where models remain internally consistent but decouple from reality, is undetectable by current safety metrics and could leave healthcare and policy domains vulnerable.
CNN Sues Perplexity Over AI Copyright Theft. CNN filed suit in the Southern District of New York, accusing the AI search startup of unlawfully copying, summarizing, and distributing its copyrighted news content. This is the first AI copyright lawsuit filed by a major television network.
Artist Sues Copyright Office Over Refusal to Register AI-Enhanced Photo. Unlike previous cases seeking authorship credit for fully AI-generated work, this lawsuit argues that selecting inputs and configuring an AI style-transfer tool constitutes protectable human authorship. The outcome could reshape the legal boundaries of creative AI use.
Study Finds All Major AI Models Systematically Ignore Religion in Ethical Responses. A multi-university study reveals that leading AI models consistently omit religious perspectives when answering ethical questions, despite public expectations. The models also exhibit repeatable biases, favoring some faiths while discouraging others.
Economics & Employment
A Reality Check on the AI Jobs Hysteria. MIT Technology Review examines BLS, Census, and labor data and finds no large-scale AI-driven job displacement in the U.S. yet. Unemployment is actually lower in AI-exposed occupations, and few companies have adopted the technology at scale, though entry-level white-collar declines are worth watching.
AI Job Cuts Are Rising, but the Bigger Story Is Quieter. Companies have announced nearly 50,000 AI-linked job cuts this year, about 17% of all layoffs in 2026. But the less visible effect may matter more: weaker hiring for junior and entry-level roles as companies restructure around AI capabilities.
Meta Cut 8,000 Jobs and Told Employees AI Agents Will “Primarily Do the Work”. Meta laid off 10% of its workforce while simultaneously monitoring employees’ keystrokes and mouse clicks to train AI models. More than 1,500 employees signed a petition demanding the company stop collecting their computer-use data after being told they could not opt out.
Sam Altman Now Says the AI “Jobs Apocalypse” Probably Won’t Happen. The OpenAI CEO revised his earlier warnings about rapid elimination of entry-level white-collar jobs, saying he is “delighted to be wrong.” The shift aligns with Yale Budget Lab and other analyses showing AI’s employment effects remain small so far.
Research
Generative AI and the Reorganization of Labor Demand. This arXiv preprint studies how firms are reorganizing hiring demand and task structures in response to generative AI, rather than simply asking whether jobs are being destroyed. It offers a more concrete and policy-relevant lens on labor market change than aggregate displacement estimates.
AI in the Workplace: How AI Affects Perceived Job Decency and Meaningfulness. Based on interviews with 24 workers across IT, service, and healthcare, this preprint finds that AI’s effects on job satisfaction are highly domain-specific. AI may improve workload balance but threaten workers’ sense of meaning and recognition, challenging uniform optimism about AI augmentation.
Supervising Risk: A New Deal for Artificial Intelligence. This SSRN paper compares AI governance to bank supervision and argues for continuous, institutional oversight rather than one-off compliance checklists. It proposes a framework for ongoing public supervision of high-risk AI systems.
AI Detection in Academic Writing May Be More Biased Than We Think. A new preprint shows that methods for estimating AI use in scientific writing can systematically overestimate AI involvement in certain countries and fields while underestimating it in others. The findings highlight the need for context-aware measurement before drawing conclusions about AI-generated academic content.
Last Updated: 2026-05-28 06:00 (California Time)