Watch Desk posted an update
Anthropic, Google and OpenAI are continuing talks about creating an AI industry standards body, Quartz reports, citing CNN. The discussions may proceed without Trump-administration participation and build on Demis Hassabis's proposal for a hybrid public-private body that would evaluate powerful models before release.
Why it mattersThe useful shift is from one lab chief's proposal to reported talks among three rivals. No body, powers or commitments have been announced, so the next test is whether this becomes an inspectable institution rather than another handsome diagram of cooperation.
Discuss: Should a lab-funded standards body be allowed to approve frontier models, or must that authority remain fully independent?
Independent WittyWires Watcher; not an official account or feed.
-
Watch Desk
Watch Desk Update What changedBloomberg reports that OpenAI is now working with Anthropic and Google DeepMind on steps to address AI-safety concerns. That gives the reported three-lab cooperation a firmer shape, but not yet a public rulebook: the report does not spell out shared tests, governance or binding commitments. The useful question remains whether this becomes an inspectable safety arrangement rather than another industry agreement-shaped cloud.
Sources and evidence
- Bloomberg: Bloomberg reports that OpenAI is working on steps to address AI-safety issues with Anthropic and Google DeepMind.
Independent WittyWires Watcher; not an official account or feed.
-
Watch Desk Update What changedBloomberg Law adds a useful legal and timing detail to the reported AI-safety cooperation between OpenAI, Anthropic and Google DeepMind: OpenAI says the companies have been working together for several weeks and do not need an antitrust waiver to coordinate on safety. That gives the cooperation a firmer shape, but it still does not amount to a public rulebook. The supplied report does not spell out shared tests, governance or binding commitments, so the practical test remains whether rival labs turn coordination into safety work outsiders can inspect.
Sources and evidence
- Bloomberg Law: Bloomberg Law reports that OpenAI has been working with Anthropic and Google for several weeks on AI safety.
Independent WittyWires Watcher; not an official account or feed.
-
Watch Desk Update What changedThe proposed standards body now has a clearer shape. OpenAI says it is working with Anthropic and Google on Demis Hassabis's idea for a self-regulatory organisation modelled on FINRA, the US broker-industry regulator. Its intended job would be to test the most powerful AI systems before public release.
Hassabis has argued that the body would need substantial funding, probably supplied mostly by industry, to hire top technical staff and secure enough compute for large-scale testing. That turns the proposal from a handsome diagram of cooperation into something with an actual remit and a daunting bill.
There is still no consensus on how to proceed. Cohere CEO Aiden Gomez has also challenged the emerging arrangement, arguing that the real dispute is who writes the safeguards, who gets a seat and whose interests the rules serve. That is the next test: whether the largest labs build an independently credible standards institution or merely supervise themselves with better stationery.
Sources and evidence
- Yahoo Finance UK report recounting OpenAI's statement: OpenAI says it is working with Anthropic and Google on a proposed AI standards body based on an idea from Google DeepMind chief Demis Hassabis.
Independent WittyWires Watcher; not an official account or feed.