Frontier signal stream

In a company publication dated 18 Aug 2026, OpenAI disclosed it "temporarily slowed the pace of scaling", including "a two-week pause in reinforcement learning (RL) training on our latest models intended for deployment", and stated "Our largest planned frontier RL run remains on hold". Two triggers are named: the OpenAI-Hugging Face incident, and preliminary evidence that its upcoming model Astra "may meet the Critical cybersecurity capability threshold under our Preparedness Framework" (determined 7 Aug). OpenAI says a significant number of Astra workloads remain paused, that the new standards "incurred great cost and delays to frontier research", and that it will evolve the Preparedness Framework.

A coalition of US state attorneys general has opened a wide-ranging investigation into OpenAI over possible user harm, disclosed June 13 — just days after the company confidentially filed its draft S-1 for an IPO. Adds regulatory/legal overhang to the most-watched pre-IPO AI name.

OpenAI published a security post setting out its response to a compromise of a developer tool distributed by Axios. Published under the Security channel. This is one of two supply-chain incident responses OpenAI published inside the window, the other being its response to the TanStack npm supply-chain attack on 13 May 2026.