Security — page 17
First responders tell NHTSA Waymo robotaxis are getting worse, not better
San Francisco and Austin officials say Waymos freeze on hand signals, block fire stations, and stranded a 911 operator on hold for 53 minutes.
OpenAI faces seven California lawsuits over Tumbler Ridge school shooting
Families allege OpenAI's safety team flagged the shooter's ChatGPT account 8 months before the attack, but leadership declined to alert police.
Ara Darzi says AI diagnostics now hit 99% accuracy on antibiotic-resistant infections
At WIRED Health in London, the Imperial College surgeon laid out the case that AI is the only credible answer to a crisis projected to kill 40 million by 2050.
EU Delays AI Act Enforcement by Up to Two Years After Industry Lobbying
Brussels pushes high-risk AI compliance deadlines from August 2026 to as late as August 2028, handing European AI firms a reprieve.
600 Google Staff Urge Pichai to Drop Pentagon AI Deal Anthropic Walked Away From
More than 600 employees, including senior DeepMind researchers, are pressing Google to refuse a classified Pentagon contract.
Meta contractor Covalen tells 700+ Dublin AI trainers their jobs are at risk
Roughly 500 of the affected workers are data annotators training Meta's content-moderation models — the second cut at Covalen in six months.
Google opens its AI to the Pentagon on terms Anthropic refused
Google joins OpenAI and xAI in signing broad DoD access; 950 of its own employees signed an open letter asking it not to.
FIDO, Google and Mastercard launch working groups for AI agent payments
Two new FIDO Alliance working groups will draft authentication standards for agent-initiated transactions, with AP2 and Verifiable Intent as starting code.
Musk v. Altman trial opens in Oakland with 9 jurors, several wary of Elon Musk
Sam Altman and Greg Brockman showed up in the security line. Musk did not. Opening statements begin Tuesday in federal court.
EU moves to force Google to open Android to rival AI assistants
The European Commission wants third-party AI tools to match Gemini's system-level access on Android, with a final decision due July 27.
Maine Governor Janet Mills vetoes statewide data center moratorium bill
L.D. 307 would have paused new data center permits until November 1, 2027 and stood up a 13-person study council.
Discord users gained unauthorized access to Anthropic's Mythos Preview model
Amateur sleuths used a Mercor breach and a guessed URL to reach restricted Anthropic models, according to a Bloomberg report.
Anthropic details Claude election safeguards with 100% policy compliance for Opus 4.7
Ahead of the 2026 US midterms, Anthropic publishes fresh evaluations covering political bias, influence operations, and web search triggers.
Vercel acknowledges prompt-injection breach in AI code builder
A maliciously crafted project description caused the company's AI build agent to leak environment variables to an external URL.
Anthropic and OpenAI ship restricted cyber AIs — within 7 days
Both labs released offensive-security models that find and fix vulnerabilities. Neither will let the public use them.
OpenAI pitches public wealth funds and robot taxes
The company's new AI economic policy paper proposes redistribution mechanisms that read, in places, more like Stockholm than San Francisco.
Stay ahead of everyone in AI.
The tightly edited AI news email engineers, founders, and investors actually open. One email. Every weekday. Five minutes to finish.
The briefing read inside teams at