California Orders AI Safety Laws by 2027 After OpenAI, Anthropic Incidents

Key Takeaways

• California Governor Newsom signed an AI oversight executive order calling for kill switches and independent safety audits for frontier models, with recommendations due by mid-2027.
• The order follows AI incidents involving OpenAI and Anthropic agents, and experts must report recommendations by November 16.
• EFF supports the order but warns government kill switches could be misused to suppress free speech and urges focus on biased algorithms.
• President Trump dismisses AI safety concerns as a hoax, while Americans broadly demand guardrails on the technology.
• Anthropic CEO Dario Amodei and Anthropic's Jack Clark are among those urging a safety framework amid growing fears about AI's impact.
• Democrats are pushing for a clear AI strategy through the Tomorrow Together initiative led by Rob Flaherty, with experts warning of catastrophic outcomes if AI remains unregulated.
• Kara Swisher says the real danger from AI comes from the people building it, not the technology itself, criticizing Silicon Valley's repeated failure to take precautions.
• Congress heard that AI military targeting may have accuracy closer to 50%, not 90%, and the Pentagon's Civilian Protection Center of Excellence has been weakened.
• ChatGPT Images 2.5 introduces point-and-sketch editing, letting users edit specific image parts and sketch shapes as instructions instead of writing long descriptions.
• A Kaspersky study warns viral AI avatars expose permanent biometric data, with 27% of Latin American users not checking app permissions before installing.

Newsom signs AI oversight order with kill switch plan

California Gov. Gavin Newsom signed an executive order to speed up AI oversight. The order directs agencies to recommend stronger safety laws by mid-2027. Proposals include independent safety audits and emergency kill switches for frontier AI models. Newsom criticized the federal government for failing to regulate AI. He wants California to set the national standard for AI safety.

Newsom orders new AI safety plans after vetoing tough rules

Gov. Gavin Newsom ordered California agencies to develop new AI safety regulations. This comes after he vetoed stricter rules two years ago. The move follows AI incidents involving OpenAI and Anthropic agents. Experts must report recommendations by November 16. Newsom is considering a special legislative session on AI safety.

EFF supports Newsom AI order but urges broader focus

The Electronic Frontier Foundation praised Newsom's AI executive order. EFF said it starts a needed conversation about AI harms. The group urged focus on current issues like biased algorithms. EFF supports third-party investigations and loss-of-control reporting. It also warned that government kill switches could be misused to suppress free speech.

Newsom issues AI safety order as fears grow nationwide

California Gov. Gavin Newsom issued an executive order on AI safety. The order calls for experts to suggest stronger state AI laws within two months. Possible new rules include kill switches and third-party safety plans. Newsom said Washington has failed to regulate AI. He stated California should lead with the strongest AI framework in the nation.

Gulf states must help shape AI governance, opinion argues

An opinion piece argues Gulf states should engage in AI governance debates. The author notes that anxiety about AI is real, especially from insiders like Anthropic CEO Dario Amodei. Immediate concerns include job loss in white-collar fields. The IMF warns of wage inequality from AI adoption. The article says Gulf nations have strong AI investments and a voice in shaping rules.

Trump opposes AI safety as Americans demand rules

President Trump calls AI safety concerns a hoax and favors unregulated growth. Americans want guardrails on the technology. AI leaders like Anthropic's Jack Clark urge a safety framework. The push for rules is urgent because AI could change history. Trump's stance is driven by economic and political benefits.

Democrats face a tough challenge on AI policy

Democrats are urged to create a clear AI strategy. Rob Flaherty leads the Tomorrow Together initiative to push for AI regulation. Experts warn that unregulated AI could lead to catastrophic outcomes. Voters are paying attention to the issue. Democrats have an incentive to act before Republicans do.

Viral AI avatars expose your face data forever

Kaspersky warns that making viral AI avatars exposes biometric data that cannot be changed. Photos can reveal documents or locations that help thieves steal identities. 27% of users in Latin America do not check app permissions before installing. Kaspersky advises using trusted apps and reviewing photos carefully before uploading.

ChatGPT Images 2.5 adds point and sketch editing

ChatGPT Images 2.5 lets users point at specific parts of an image to edit it. Users can sketch shapes as instructions instead of writing long descriptions. The tool keeps earlier fixes intact when making new edits. It helps with tasks like adjusting shadows or adding branding. The feature streamlines image editing workflows.

Digital twins struggle in real world settings

Digital twins are virtual copies of real systems used in industries like healthcare and manufacturing. A Columbia Business School study found they work well in labs but fail in real-world situations. The problem is poor data and difficulty capturing human behavior. Researchers suggest making twins more flexible and combining them with machine learning. Better design could make them useful.

Broadcom's AI Push Drives 123% Revenue Growth vs Applied Materials

Broadcom and Applied Materials are both AI-related companies but their sales growth is very different. Broadcom's revenue grew 123% over two years thanks to heavy AI investments. Applied Materials grew only 10% as it focuses on the slower-moving semiconductor business. The comparison shows how AI focus can dramatically boost a company's sales.

Kara Swisher Says People Building AI Pose the Real Danger

Tech journalist Kara Swisher says AI itself will not destroy humanity but the people building it might. She spoke at The Atlantic Festival and said Silicon Valley keeps making the same mistakes with new technology. Swisher said tech leaders are very smart but often fail to take proper precautions. She also discussed who might win the AI race and how Americans should respond.

Congress Told AI Is Already Causing Civilian Deaths

Lawmakers heard at a congressional hearing that AI is already being used by the US military to target enemies, with civilians getting hurt. The Pentagon says AI could reduce civilian harm, but experts dispute that claim. Steven Feldstein of the Carnegie Endowment said AI targeting accuracy may be closer to 50%, not 90%. The Pentagon's Civilian Protection Center of Excellence, set up in 2023, has been weakened, witnesses said.

Sources

• Gavin Newsom Moves To Lasso AI With Oversight Executive Order
• Newsom Orders California Agencies to Develop New AI Safety Plans After Rejecting Tougher Law
• EFF Statement on California Governor's Executive Order on AI
• California Gov. Newsom issues executive order to rein in AI 'before it's too late'
• Big Tech need not give artificial intelligence a bad name
• Americans want AI safety, Trump doesn't. Guess who's winning
• The ‘Political Disaster’ Looming for Democrats on AI
• Creating a viral avatar with artificial intelligence exposes unchangeable biometric data
• AI Tool of the Week: this ChatGPT feature lets you point, sketch and edit images
• Digital twins stumble on surveys but may pay off in the field: Here’s why
• Applied Materials vs. Broadcom: Comparing Sales Growth Trajectories for These Artificial Intelligence Companies
• Kara Swisher: AI Isn’t Going to Destroy Humanity—But the People Building It Might
• Congress told AI is already killing people