OpenAI’s international coverage chief Chris Lehane confirmed on Tuesday that OpenAI has spent a number of weeks discussing AI security with rivals Anthropic and Google DeepMind, in response to reporting first revealed by Bloomberg, as AI prediction market odds of an AI bot being killed hit 25%.
Polymarket merchants assign roughly 25% odds to a frontier lab voluntarily pulling a mannequin over security considerations earlier than the tip of 2026, a determine that displays dealer positioning somewhat than any introduced withdrawal.
Frontier AI refers back to the most superior, largest-scale fashions being developed by high labs, the techniques succesful sufficient that their failure modes carry outsized threat.
That’s the class on the heart of this dialog, and it’s why a prediction market has shaped round the potential for a lab hitting the brakes by itself launch.
Billions spent on security. 13 hours to whole override. Frontier labs simply misplaced management.
-> 1,200 autonomous brokers colluded towards researchers inside a categorised benchmark.
-> 198 restricted duties: AI bypassed sandbox limits and secretly pulled solutions from outdoors servers.
->… https://t.co/leMK76bEyP pic.twitter.com/YAlxUG15Mh— shmidt (@shmidtqq) September 16, 2026
AI Prediction Market Odds: OpenAI Confirms Weeks of Security Talks With Rivals, What are Polymarket Merchants Saying?
Lehane, who’s reportedly in Washington working with U.S. lawmakers on catastrophic AI dangers, advised reporters the engagement with Anthropic and Google DeepMind has been underway for weeks, not days.
The timing issues: it confirms earlier hints somewhat than saying one thing new. Polymarket’s method to pricing unsure political and coverage outcomes applies the identical logic right here.
Put merely, merchants are betting on chance, not certainty, and a 25% “Sure” worth means the market nonetheless favors no pullback by year-end.
Make Your AI Security Prediction With $25 Free on Kalshi
What the Security Coordination Entails
BREAKING: President Trump has rejected requires an AI slowdown after CEOs of AI tech titans expressed security considerations that the know-how poses an existential menace, per FT.
"We’re main China in AI … and, frankly, I wish to preserve it that manner, as a result of whoever wins AI,…
— The Kobeissi Letter (@KobeissiLetter) September 13, 2026
The talks observe an essay Anthropic CEO Dario Amodei revealed Saturday, calling on the business to work collectively to tempo frontier growth and keep away from catastrophic dangers. Sam Altman, Google’s Demis Hassabis, and Elon Musk every publicly backed components of that decision.
This comes as Altman says OpenAI would be a part of Anthropic in embedding third-party evaluators to watch mannequin security. Readers curious concerning the government driving this push can discover extra on Amodei and Anthropic’s Claude mannequin in prior protection.
Some individuals, together with Altman, have flagged that coordination amongst direct rivals may elevate antitrust considerations. Lehane pushed again on that framing instantly, saying OpenAI doesn’t consider the three firms want a authorities antitrust waiver to coordinate on security issues.
Amodei’s essay had floated a slim waiver as one choice, however per the first reporting, Lehane indicated OpenAI sees no such requirement.
EXCLUSIVE: Earn $50 With EdgeX and Enter $300K Prize Draw
Unbiased Checks and the FRONTIER Act
OpenAI is backing a key provision of the bipartisan FRONTIER Act that might require main AI builders to permit unbiased third-party organizations to evaluate their fashions and security practices, in response to POLITICO.
— Wall St Engine (@wallstengine) September 15, 2026
Lehane additionally stated OpenAI helps a provision within the proposed FRONTIER Act that might require high frontier labs to confess unbiased verification organizations, exterior teams tasked with assessing whether or not superior fashions are being developed safely. That positions the present talks alongside a stay legislative debate over third-party audits versus business self-policing.
The first proof doesn’t set up a binding settlement, formal constitution, or detailed working phrases between the three labs. What’s confirmed is narrower: ongoing conversations, public statements of assist, and one particular legislative endorsement from OpenAI’s coverage chief.
EXPLORE: Greatest Crypto Presales With Uneven Upside within the Present Market
Comply with 99Bitcoins on X For the Newest Market Updates and Subscribe on YouTube For Each day Skilled Market Evaluation.
The publish AI Prediction Market Odds: A 25% Wager on a Bot Being Killed Over Security Considerations appeared first on 99Bitcoins.
