The U.S. administration lifted export controls on Anthropic's most advanced language model, Fable 5, the latest turn in several weeks of unusually direct federal involvement in frontier-AI releases. OpenAI, maker of ChatGPT, is now facing a similar release problem: GPT-5.6 is starting with “trusted partners” whose participation has been shared with the government, while OpenAI has reportedly discussed giving the administration a 5% stake in the company.
California is making Anthropic’s Claude available to state agencies and eligible local governments at a discount. (Business Insider)
Two Taiwan-based Super Micro employees were detained in an Nvidia-chip export probe, though the company says it is not a target of the investigation. (Reuters)
Hackers may have breached a DHS network used to share sensitive security information. (Nextgov/FCW)
Nvidia and Palantir are partnering to deploy "sovereign AI" to U.S. agencies and enterprises. (eeNews Europe)
The FTC is looking at whether hidden AI model tuning should be treated as consumer deception. (GovInfoSecurity)
Federal prosecutors in Los Angeles used a defendant’s ChatGPT history as evidence in an arson case tied to the 2025 Palisades Fire. (TechDogs)
Axon’s CEO is pitching AI as the next big operating system for police departments. (Wall Street Journal)
Apple released security updates ahead of schedule, citing AI that shortens the time to exploit bugs. (ITPro)
