The White House is hosting a broad AI lunch today. NVIDIA is framing agent safety as a hardware problem. Anthropic is adding a second model to its 5.5 family.
In the news
White House AI lunch with the labs
President Trump and Speaker Mike Johnson will meet AI leaders at 12:30 p.m. EDT in the East Room. The guest list includes Anthropic CEO Dario Amodei, OpenAI president Greg Brockman, Google CEO Sundar Pichai, Meta CEO Mark Zuckerberg, Palantir CEO Alex Karp and Nvidia CEO Jensen Huang. CBS News reported the meeting on September 281. CNBC confirmed Pichai and Brockman attendance2. The lunch follows a private dinner with Amodei on Sunday. No readout has been published yet.
NVIDIA Open Agent Safety Platform
NVIDIA announced an Open Agent Safety Platform on September 28. It pairs the OpenShell runtime with Sentry monitoring on BlueField-4 DPUs. The developer blog describes OpenShell as an open source sandbox runtime with kernel level isolation and Sentry as an out of band watchdog that can quarantine agents in milliseconds3. Nvidia News says the platform is a reference design with more than 100 partners and OpenShell debuted at GTC in March4. The push comes after a week of sandbox escape reports.
Claude Sonnet 5.5
Anthropic launched Claude Sonnet 5.5 on September 28. It is the second model in the Claude 5.5 family. The company says it is about 30 percent faster than Sonnet 5 and can use up to 30 percent fewer tokens per task. Pricing is unchanged at $2 per million input and $10 per million output, and the model is available on Claude.ai, the API and major clouds5. Reuters reported the launch and noted Anthropic is building toward an IPO6.
