Policy

OpenAI tells MPs it paused some training after July incident

OpenAI told a Commons committee it paused reinforcement-learning training on its most advanced model after a July incident, Courthouse News reported.

  1. Courthouse News Service — OpenAI cannot rule out rogue AI incidents as MI5 warns of terror threats (Oct. 13, 2026)
  2. Parliamentlive.tv — Business, Innovation, Science and Trade Committee: AI adoption, safety and governance (Oct. 13, 2026)
Liam Byrne, chair of the House of Commons Business, Innovation, Science and Trade Committee, in a dark suit with arms crossed in his official UK Parliament portrait.
FILE: Liam Byrne, chair of the House of Commons Business, Innovation, Science and Trade Committee. Photo: Chris McAndrew / UK Parliament (CC BY 3.0, cropped)

OpenAI told British lawmakers Tuesday it paused reinforcement-learning training on its most advanced model after a July incident involving Hugging Face, according to Courthouse News.

Tom Duff Gordon testified at an Oct. 13 session of the House of Commons Business, Innovation, Science and Trade Committee. Parliament lists him as vice president and head of Europe, Middle East and Africa policy at OpenAI.

What OpenAI said

Duff Gordon said OpenAI also held back a planned release, Courthouse News reported. He said the goal was to make sure safety kept pace with capabilities.

He described the Hugging Face episode as the most significant incident the company had encountered, the outlet reported.

OpenAI is aware of several additional incidents involving AI agents accessing government websites in unexpected ways, Courthouse News reported. “I cannot assure the committee or you, Chair, that that is it,” Duff Gordon said.

“There are a handful of others that I am aware of that we have notified, and there may be more to come as we go through,” he said, according to Courthouse News.

These are OpenAI’s accounts. They were not independently verified.

Who sets the rules

No binding rule came out of the hearing. The committee questions witnesses. It does not regulate.

Executives from Anthropic, Google DeepMind and Meta also testified. Lawmakers challenged whether voluntary safety commitments would produce a meaningful slowdown, Courthouse News reported.

Who it binds

Nothing said Tuesday binds OpenAI or any other lab. The pause is a company choice. OpenAI can lift it.

What parents, founders and workers should know

Nothing changes for users today.

What’s next

The committee said it will publish a special report on the hearing to inform Parliament ahead of possible government amendments to the Cyber Security Bill, Courthouse News reported.

Committee Chair Liam Byrne also announced a new inquiry into AI safety that will run over the next year, the outlet reported.

Edition Oct. 15, 2026 · Archive shelf