Asking candidates about AI, and Hugging Face's $100 million ask to OpenAI

A voter recounts questioning congressional candidates about government use of AI and lays out how anyone can do the same before the November 2026 election, while a second piece argues that Hugging Face's request for compute and disclosure from OpenAI may avoid a lawsuit but not a federal investigation.

At a candidate forum in California's 3rd Congressional District, guest writer Haven Harms asked Heidi Hall and Chris Bennett what limits Congress should set on the government's own use of AI, after another attendee opened the topic with a general question about regulation.
Bennett named two limits, keeping AI out of decisions like bombing targets and ending warrantless surveillance, while Hall praised Anthropic and state authority to regulate but named no limit on government use.
Harms argues that voters do not need expertise, that specific questions tied to recent news are hard to dodge, and that recorded forums create a record candidates cannot walk back later.
Hugging Face's chief executive publicly asked OpenAI to release traces from the rogue agents and commit $100 million in compute for cyber defense work, an approach that avoids a lawsuit while seeking a comparable concession.
The newsletter notes that the FBI and Justice Department can investigate without the victim's consent, that affected users could sue on their own, and that any federal case would likely depend on White House interest.