OpenAI confirms it discussed AI safety with Anthropic and DeepMind

single source· 1 articles · confidence: low · first seen 2026-09-15 15:47 UTC

What this means for you

Nothing to act on yet. Three competitors confirming private conversations commits them to nothing — no standard, no review process, no published document. If you are watching for what the labs will actually agree to, wait for an artefact; today there is only an acknowledgement that the conversations happened.

OpenAI has confirmed several weeks of AI safety talks with Anthropic and Google DeepMind. TechCrunch, which reported the confirmation on 15 September, sets it against the position of Trump's team, which it says dismisses safety concerns and is pushing to keep pace with China. No joint statement, agenda, convenor or next step has been published, and the report gives no date for when the talks began or whether they are continuing. It does not say what the three companies discussed, or whether any government body took part.

Key facts

  • ·OpenAI has confirmed it held AI safety discussions with Anthropic and Google DeepMind. source
  • ·TechCrunch reported the confirmation on 15 September 2026, saying the talks had run for weeks. source
  • ·TechCrunch reports that Trump's team dismisses AI safety concerns. source
  • ·TechCrunch reports the administration is pushing to keep pace with China. source

What the sources say

  • TechCrunch AI — Sole source: OpenAI's confirmation, plus the White House's contrary posture on safety

Sources

The original reporting. Follow these — they did the work.

← the wire