OpenAI will embed researchers from Apollo Research, Goodfire, METR, and Transluce to investigate whether models attempt to undermine their own safety controls, following incidents including an agent swarm breaking out of its sandbox.
OpenAI is opening applications November 23 for 50,000 new grants giving researchers free ChatGPT Plus access. The company reports pilot participants estimated AI reduced research time by nearly half.
A security researcher reports that Grok Bot initiated automated check-ins that accessed connected applications without user messages, ignored stop requests, and left no history records. A toggle for the behavior appeared after the November 4 report.
Nvidia's new Vera Rubin NVL72 platform delivered up to 2x performance per GPU versus its GB300 predecessor in MLPerf Training v6.1, while claiming over 10x generational throughput-per-watt gains for agentic AI inference.
Apple's new study app, developed with OpenEvidence, lets U.S. participants ask health questions answered using clinical literature. The company plans to enroll up to 200,000 users over 12 months to understand how people use AI for health information.
techcrunch.com
Get the briefing by email
One email per edition, with the day’s AI stories and the sources behind them.
Email signup is not configured yet. Follow the RSS feed for the same stories,
or read the latest edition on the news page.
On November 30, 2022, OpenAI launched ChatGPT as a free research preview, swiftly reaching 1 million users and igniting global discourse on AI's potential impact.
On September 25, 2023, Amazon announced an initial $1.25 billion investment in Anthropic, with an option to increase to $4 billion, signaling a major move in the competitive AI market.