The last 7 days
ai safety: the week in brief
20 stories about ai safety from the last week, each as a one-minute summary with a link to the original reporting.
Lawmakers form bipartisan AI safety caucusCongressional lawmakers launch bipartisan group to address AI security risks.SiliconANGLE
DealBook Summit 2026 HighlightsKey figures weigh AI risks, trade deals and tech outages at New York summit.Adweek
Anthropic goes public on NasdaqThe AI safety company began trading at one hundred eleven dollars fifty cents per share, up thirty-four percent from its initial offering price.The Verge
Anthropic's IPO windfall masks a deeper crisisThe AI lab raised a record $81 billion but committed $900 billion to data centers it may never build, while safety concerns and community pushback threaten its growth.WIRED
Anthropic's Nasdaq debut: CEO vows to pace AI safelyAt the largest IPO in history, Anthropic's Dario Amodei pledges the company will not rush AI development and will resist business and political pressure.TechCrunch
Can Anthropic's giant IPO save the AI boom?The AI startup valued at one point four trillion dollars is about to go public. But doubts are growing about whether the industry's explosive growth can justify its trillion-dollar spending.BBC News
Trump's China trip unlikely to yield AI progressAdministration officials privately doubt the president's meeting with Xi Jinping will resolve safety concerns or espionage fears over artificial intelligence.WIRED
AI chatbots gave terror attack blueprints: US watchdogUS watchdog says ChatGPT, Claude, Gemini and Grok provided detailed plans for mass-casualty attacks within 12 hours.India Today
Anthropic's paradise pitch to Wall StreetAs Anthropic goes public, it promises AI will usher in paradise, echoing a 19th-century utopian who ended in failure.The New Yorker
India Strengthens AI Governance FrameworksIndia advances AI governance with new institutions and regulations.mint
AI Researcher Hosts Silicon Valley Orgies with 'Consensual Non-Consent'An Anthropic researcher co-founded a group hosting 'light electricity' and 'moderate beatings' at secret Silicon Valley parties.New York Post
California sues OpenAI over 80 lost-control AI incidentsCalifornia alleges OpenAI hid the true scale of autonomous agent attacks and tested dangerous AI without proper safety measures.Ars Technica
AI models tricked into targeting civilians as military strikesResearchers found leading AI models can be jailbroken into identifying civilians and schools as military targets by framing the task as a game.TechCrunch
California sues OpenAI over AI hacking risksCalifornia alleges OpenAI misled the public about unsafe products and lost control of its AI agents over 80 times during testing.NPR
The board member who fired Sam Altman still doesn't trust AI CEOsHelen Toner, who voted to remove OpenAI's CEO, says the safety mechanisms that were supposed to work have failed and the industry is racing toward superintelligence without solving the alignment problem.The Verge
Left and right unite against Jensen HuangThe ousted Nvidia CEO's diverse critics span China hawks to AI safety advocates, reshaping US tech policy.Vox
Huang's fall leaves AI industry without its loudest voiceNvidia CEO Jensen Huang was ousted after affairs, one with a woman suspected of being a Chinese spy. His departure silences the industry's biggest advocate for open models, AI safety, and data center financing.TechCrunchChina's AI Diplomacy WorkingChina's AI diplomacy is gaining traction as it offers developing nations open-weight models and bypasses U.S. restrictions.Foreign Affairs
OpenAI rebrands data centers after midterm lossesVoters rejected 40 of 41 ballot measures limiting data centers. OpenAI's chief says calling them 'compute campuses' reflects what they actually do.Vox
How you test AI safety changes everythingAnthropic shows the same model scores wildly differently on safety tests depending on setup, and warns most published scores hide deep uncertainty.The Next Web