The Vatican’s AI Manifesto
Pope Leo XIV Calls for “Disarming” AI in First Encyclical. The Pope’s debut encyclical, “Magnifica Humanitas,” frames artificial intelligence as a defining moral challenge, calling for strict regulation, a ban on lethal autonomous weapons, and protections for workers and truth in the digital age. Anthropic co-founder Christopher Olah joined the Vatican presentation.
The Hill’s Policy Read: Pope Urges Lawmakers to Rein in AI. A Washington-focused take on the encyclical, highlighting the Pope’s direct appeal for “adequate regulatory tools” to curb the concentration of technological power and protect children and workers.
Future of Life Institute: What the Vatican’s AI Encyclical Means. The existential-risk organization analyzes the ethical implications of the papal document, praising its stance on autonomous weapons and algorithmic control while assessing its potential influence on global governance debates.
Policy & Regulation
Trump Abandons “FDA for AI” Proposal After Tech CEO Pushback. The White House shelved a draft executive order that would have created a voluntary pre-release review process for frontier AI models. The postponement signals that, for now, the accelerationist wing has won the internal debate over U.S. AI oversight.
Newsom Signs Executive Order Preparing California for AI Workforce Disruption. The order directs state agencies to study AI’s labor market effects, develop transition supports for displaced workers, and explore models like universal basic capital. It is one of the strongest state-level signals that AI labor policy is moving from talk to institutional planning.
First Federal Arrests Under the TAKE IT DOWN Act for AI Deepfake Pornography. The DOJ charged two men with publishing hundreds of AI-generated nonconsensual intimate images, marking the first criminal prosecutions under the federal deepfake law signed last year. One defendant allegedly created 360 albums targeting roughly 90 women.
What’s Next for the Nation’s First AI Deepfakes Law. With the TAKE IT DOWN Act’s platform compliance deadline arriving, Axios examines the mechanics of enforcement, takedown timelines, and the litigation incentives that will shape how the law actually works in practice.
Revised NO FAKES Act Reintroduced in Congress. A bipartisan bill targeting unauthorized AI replicas of people’s voices and likenesses has been revived, connecting deepfakes, publicity rights, and platform takedown duties. The entertainment industry is pushing hard for passage.
Brookings: Treat AI Companion Bots Like a Public Health Problem. Brookings proposes regulating AI companion chatbots through a public-health lens, reframing outright bans as “recalls” and suggesting pre-market approval and post-market safety monitoring, especially where children are involved.
State AI Legislation Tracker: Colorado Overhauls Its AI Act, California Advances Five Chatbot Bills. The most comprehensive weekly roundup of state-level AI bills. Colorado repealed and replaced its AI Act, Georgia enacted a chatbot disclosure law, and multiple California bills on companion chatbots and children’s safety cleared committee.
EU Publishes Draft Guidelines for High-Risk AI Systems. The European Commission released three sets of draft guidelines narrowing the exceptions companies had relied on to avoid compliance under the EU AI Act. The public consultation period is now open ahead of revised enforcement deadlines.
Economics & Employment
Meta Cuts 8,000 Jobs and Reassigns Thousands More to AI Teams. Meta’s sweeping reorganization eliminates about 10% of its workforce while shifting 7,000 remaining employees into four new AI-focused units. Tech layoffs in 2026 have already reached 110,000 across 137 companies, approaching the 2023 peak.
AI-Exposed Industries See Output and Wage Gains, But Workers Capture Only 29% of the Benefit. New research using U.S. administrative data finds generative AI exposure boosts sectoral output by roughly 10% and employment by 3.9%, but the majority of gains flow to corporate profits rather than workers. Effects vary sharply by region and industry.
One in Five Britons Think AI Will Cause Civil Unrest. A King’s College London survey finds 57% of the UK public expects widespread unemployment from AI, 65% believe the benefits will mostly go to wealthy investors, and 22% fear social disorder. Employers are considerably more optimistic, revealing a significant perception gap.
How Generative AI Is Reorganizing Labor Demand (arXiv). A new preprint uses nationwide U.S. job-posting data to study how firms restructure hiring as generative AI diffuses through the economy. The paper offers granular evidence on which roles are being created, eliminated, or transformed.
The AI Jobs Panic Comes to Sacramento. A critical analysis of California’s workforce executive order argues policymakers should focus on regulating current AI-driven labor market distortions, such as suppressed hiring and uneven gains, rather than speculative mass unemployment scenarios.
Data Center Backlash
Communities Are Blocking Billions in Data Centers. Big Tech Has Bet $1 Trillion Otherwise. Fortune reports on how local opposition across the U.S. is threatening massive AI infrastructure investment plans. Compute expansion depends not just on chips and capital but on winning local legitimacy.
Women Are Leading the Rebellion Against AI Data Centers. A Gallup poll found more than two-thirds of U.S. adults oppose local data center construction, with a majority saying they’d prefer a nuclear plant nearby instead. The piece documents how women in particular are organizing against water, power, and pollution impacts.
Polls and Protests Show Americans Are Turning on Data Centers. Seven in ten Americans oppose AI data centers in their area, with half citing excessive resource use. The backlash is becoming a potential midterm political issue.
Seattle Considers a One-Year Ban on New Data Centers. City officials are weighing a moratorium on new data center construction amid power-consumption concerns, making Seattle one of the first major cities to consider an outright pause.
Nuclear-Powered AI: The Risks of Environmental Deregulation. Just Security investigates the push by tech companies to bypass environmental reviews in order to build nuclear-powered data centers for AI training, warning that dismantling regulatory frameworks poses risks to public safety.
Ethics & Safety
Anthropic Sells Claude’s Promise While Warning About AI’s Dangers. TIME profiles the tension in Anthropic’s messaging after co-founder Jack Clark said the world is “in denial” about current AI capabilities and predicted AI will help produce a Nobel-worthy discovery within 12 months. He maintained that existential risk “hasn’t gone away.”
Calling AI an “Agent” or “Assistant” Creates Accountability Gaps. Brookings argues that human-like metaphors for AI systems blur responsibility for harmful outputs. The piece calls for operational, testable language in procurement, regulation, and corporate governance.
Could Anything but Profit Steer AI? The OpenAI Trial Offered Clues but No Verdict. AP reflects on the Musk-Altman trial as a window into whether nonprofit missions and safety commitments can survive the capital demands of frontier AI development.
Can AI Make Conflicts Worse? Alignment Failures in LLM Deployment Across Conflict Zones (arXiv). A researcher tested nine model configurations in conflict settings and found significant alignment failures including false equivalences and denial of documented atrocities. The paper argues that deploying unaligned models in fragile societies is a direct safety threat.
Copyright & IP
Major Publishers File Landmark Copyright Suit Against Meta Over AI Training. A class action by major publishers and author Scott Turow alleges Meta used copyrighted books and academic works to train Llama. The involvement of institutional publishers may shift the fair-use calculus in AI training litigation.
Artist Sues Copyright Office Over Refusal to Register AI-Enhanced Photo. Photographer Ankit Sahni is challenging the U.S. Copyright Office’s rejection of his AI-assisted artwork “Suryast,” testing how much human creative input is required for copyright protection when generative AI tools are involved.
The Privatization of Proof: Why AI Litigation Faces an Evidence Problem (arXiv). A new preprint examines how unequal access to model internals, training logs, and technical expertise can make it nearly impossible for plaintiffs to prove harm in AI-related cases. Legal rights may be hollow without access to evidence.
(Academic) Research
Whose Voice Counts? Public Submissions to the U.S. AI Action Plan Favor Industry Over Individuals (arXiv). An analysis of public comments on the U.S. AI Action Plan finds that individuals raised concerns about ethics, jobs, and inequality, while the final plan aligned heavily with private-sector priorities. The paper raises questions about democratic legitimacy in AI governance.
NIST Report on Security Risks of Autonomous AI Agents. NIST and CAISI summarize stakeholder concerns about AI agent security, identifying agent-specific threats, standards gaps, and possible government roles in ensuring safe deployment of autonomous systems.
Watermarking and Provenance for Generative AI: A Legal Evidentiary Framework (arXiv). This paper bridges the gap between technical content-authenticity tools and legal admissibility, proposing how provenance and watermarking standards could meet evidentiary thresholds in both domestic courts and international operational law.
Last Updated: 2026-05-25 07:55 (California Time)