Antitrust rules are the competition laws that stop rival companies from colluding on prices or shutting out markets. Safety coordination is different in spirit: labs talking about shared standards and limits so models do not race past controls. Ribera told POLITICO Europe she would entertain a request by leading US AI labs to cooperate on safety. Openness to discuss is not an exemption already granted.
Anthropic CEO Dario Amodei has called for common safety standards and limits on unchecked progress. Sam Altman, Elon Musk, Demis Hassabis and Satya Nadella have backed that kind of cooperation publicly. The politics of "please let us talk" now meets the legal question of how far rivals can go before Brussels says no.
TL;DR: Ribera says the EU would discuss room for AI labs to work together on safety under competition rules. Amodei pushed common standards; other big names backed the idea. A Commission spokesperson said antitrust is not what blocks safe AI and Brussels is ready if companies approach. No exemption has been granted yet.
Status note: Checked 27 September 2026. Ribera's comments are from 16 September 2026 POLITICO Europe reporting. Any later formal filing or decision is outside this article unless confirmed separately.
What Brussels actually said
Ribera's line is conditional. She would entertain a request. She did not announce that OpenAI, Anthropic, Google DeepMind or anyone else already holds a waiver. A Commission spokesperson put it more bluntly: it is not antitrust rules that stand in the way of safe AI, and the Commission is ready to engage if companies approach.
That framing matters. It tells labs the door is open for a conversation, and it rejects the idea that European competition law is already the villain blocking safety talks. It also leaves the burden on the companies to show up with a concrete ask.
Who is pushing cooperation
Amodei's public call for common safety standards and limits on unchecked progress is the industry signal this story hangs on. Altman, Musk, Hassabis and Nadella backing that direction publicly gives the ask political weight. It does not invent a signed EU regulation or a filed exemption application. Soften any claim that "the CEOs already got Brussels approval."
OpenAI policy chief Chris Lehane told Bloomberg that OpenAI is working with Anthropic and DeepMind on safety and does not believe a waiver is necessary. That is a live disagreement inside the industry: some want formal comfort from regulators; at least one major player says the work can proceed without a special carve-out.
What not to believe
Openness to discuss is not an exemption already granted. Do not write that the EU has legalised a safety cartel, waived antitrust for AI, or approved a specific joint standard. None of that is in the 16 September record. The facts stop at willingness to talk and a spokesperson ready to engage if approached.
Also do not invent the text of a future comfort letter or name which exact practices would be allowed. Those details do not exist in the brief.
Why it matters anyway
If labs can share safety methods without fearing a competition case, coordination gets easier. If they still fear that talking equals collusion, they may stay quieter than the public rhetoric suggests. Lehane's view that no waiver is needed sits next to Ribera's offer to discuss.
Readers should hold both lines at once. Brussels is open to a conversation. Nothing formal is done. Industry itself is not fully agreed on whether a special carve-out is required. That disagreement is part of the story, not a flaw in the reporting.
Sources: POLITICO Europe, 16 September 2026.