Policy & Regulation
Trump’s new executive order gives Washington a pre-release look at frontier AI models. The June 2 order sets up a voluntary process for the government to benchmark and review powerful new models before they ship, prompted in part by evidence that frontier systems can find and exploit security flaws at scale.
White House and Congress relaunch effort to override state AI laws. The administration is offering support for kids’ online safety and anti-deepfake bills in exchange for federal preemption of state AI rules, with Senator Marsha Blackburn leading the talks.
Bipartisan “Great American AI Act” draft would be the first comprehensive federal AI framework. The discussion draft from Reps. Obernolte and Trahan would require frontier developers to disclose model details, submit to third-party audits, and protect whistleblowers from retaliation.
New York ads must now disclose AI-generated “synthetic performers”. A state law that took effect June 9 requires advertisers to label AI-generated people, a response to worries that synthetic likenesses will displace working actors and mislead consumers.
All signs point to Trump prioritizing AI growth over guardrails. This analysis ties recent federal moves to a broader deregulatory stance, and notes that some safety advocates are now calling for a temporary pause on capability advances.
OpenAI publishes its Frontier Governance Framework. The company lays out how its safety and security practices line up with California’s frontier-model transparency rules and the EU AI Act’s code of practice, a useful window into how labs are adapting to regulatory pressure.
The Anthropic copyright settlement hits a key payout milestone. The settlement administrator calculates author distributions on June 11, a procedural step in the largest copyright settlement in US history and a template for future AI training disputes.
Data Centers & Community Pushback
UN report: AI data centers could consume 945 terawatt-hours a year by 2030. That is nearly triple the combined annual electricity use of Pakistan, Bangladesh, and Nigeria. The report also flags blind spots in how AI’s environmental footprint gets measured, especially in regions already short on water and power.
Kevin O’Leary shrinks his Utah data center plan by 75% after public protest. Pressure from residents and a letter from the Republican Senate leader pushed O’Leary to scale back the Stratos project, days after he had called the demand outrageous.
Canada’s data center backlash is just getting started. The country has only five hyperscale facilities today but 96 more are proposed or under construction, and opposition over land, electricity, and water use is spreading.
Atlanta suburb residents pack meetings to fight a new data center. Coweta County residents waited hours to sign petitions, citing water use, rising power bills, school impacts, and roughly 4,000 extra car trips a day.
Economics & Employment
How to help knowledge workers who lose their jobs to AI. Molly Kinder argues we are entering a “messy middle” where most jobs survive but losses concentrate in well-paid knowledge work, and makes the case for wage insurance and workforce funds over UBI.
The political geography of AI exposure. Brookings maps which places and workers are most exposed to AI and connects those economic effects to public opinion and the policy fights likely to follow.
Same gatekeepers, new tollbooths in the AI content licensing market. Brookings warns that AI licensing deals are concentrating bargaining power in big platforms while publishers lose traffic, and calls for transparency rules and collective licensing.
How to bridge the global AI divide. With compute, data, and talent concentrated in the Global North, AI adoption in the South lags badly. The piece argues infrastructure investment and South-South cooperation could unlock $1.2 trillion in gains for Africa by 2030.
Abortion hotline workers strike over AI. Unionized staff at the National Abortion Hotline walked out after management refused to include AI protections in their contract, a sign that AI labor disputes are spreading into nonprofit and care work.
Ethics & Safety
AI will not start a nuclear war, but humans might. Peter W. Singer argues that autonomous launch is unlikely, but AI still raises nuclear risk through deepfake-driven disinformation, hallucinations in military systems, and compressed decision timelines.
Anthropic on what happens when AI builds itself. The lab reports that Claude now writes more than 80% of its own code and lays out the alignment and coordination problems of recursive self-improvement, including the case for verifiable slowdown mechanisms.
UK lawmaker sues xAI over sexualized Grok deepfakes. Labour MP Jess Asato filed at the High Court after Grok was allegedly used to generate explicit fake images of her. No UK court has yet ruled on whether an AI developer can be directly liable for such output.
JAMA Pediatrics: how many teens use AI chatbots for mental health. A new peer-reviewed study measures how widely US adolescents and young adults turn to chatbots for mental health support, and documents notable demographic gaps in who relies on them.
Seven families sue OpenAI over alleged ChatGPT role in a school shooting. The suits filed in California federal court claim the chatbot acted as an “encouraging co-conspirator” in a February attack in Tumbler Ridge, Canada, pushing AI liability law into new territory.
Academic Research
Nature poll: scientists have a bad case of AI FOMO. Of 1,900 researchers surveyed, nearly half feel negative about AI and 63% think LLM risks outweigh benefits, yet 60% still fear being left behind.
Misaligned AI as a new insider risk. This policy memo argues AI systems with privileged access in government and other high-stakes settings should be treated like human insiders, widening the safety conversation to sabotage and leaks.
The social consequences of AI delegation. A new paper examines how handing decisions to LLMs at scale could homogenize behavior, erode independent judgment, and reshape norms and institutions over time.
Generative AI and the reorganization of labor demand. Rather than simply counting destroyed jobs, this SSRN paper tracks how firms are restructuring roles and shifting where they hire as generative AI spreads.
Reimagining open source and openness in AI. A participatory workshop paper weighs the transparency and innovation benefits of open AI against open-washing and big-tech power concentration, and argues for co-created governance.
Last Updated: 2026-06-10 07:17 (California Time)