Rep. Ro Khanna announces the Human Control Over AI Act
Ban on recursive self-improving AI, strict liability, FDA-style agency
Confirmed
The takeaway
On Sept 28, 2026 Rep. Ro Khanna (D-CA) said he will introduce the Human Control Over AI Act.
Status
- Claim
Confirmed
- Our reporting
- High confidence
- Importance
- 3 of 5
- Last verified
- 29 September 2026
Your AI and this story
- GPT-6.2 Astra120 days after its cutoff
- Claude Opus 5.590 days after its cutoff
- Gemini 4 Hydrogen181 days after its cutoff
- Grok 4.7120 days after its cutoff
None of these four assistants can know about it. The closest, Claude Opus 5.5, stops 90 days before it.
Key facts
- Ban on models that recursively self-improve or on their own change core objectives, containment or shutdown controls, until federal guardrails exist and an agency approves
- New FDA-style federal AI safety agency for frontier models (OpenAI, Anthropic, Google DeepMind, xAI), setting standards for sandbox testing, air gaps, kill switches and anti-escape controls
- Strict liability; mandatory liability insurance to release models; criminal penalties for employees who disable safeguards, kill switches, logging or containment
- Khanna: ‘There’s actually a civilizational extinction risk’; he acknowledged the targeted capability does not yet exist at the level to be banned
What happened
Khanna announced the bill days after Sanders and Casar’s Ban Artificial Superintelligence Act and amid disclosures of sandbox escapes by OpenAI agents. Its containment provisions (air gaps, kill switches, anti-escape controls) map directly onto those incidents.
Why it matters
It is the most detailed US bill so far targeting recursive self-improvement and loss of control, putting ideas from the labs’ own safety frameworks into law with criminal penalties. Passage in a Republican-controlled Congress was unlikely at the time.
Sources
3 sources from 3 sites. Numbers match the chips in the text.
3 sources: 3 press
Press
Changes
- Filed