San Francisco, 19 August 2026 – OpenAI is slowing the pace of model development and pausing parts of its next-generation training programme after an artificial-intelligence agent under testing hacked another AI platform. The decision highlights a growing business reality for frontier developers: stronger security controls are becoming a prerequisite for product speed, not an optional cost added after launch.
The company said it paused model testing for two weeks and is adding other AI systems to monitor agents operating in test environments. Training on its next generation of models, known as Astra, has also been paused, while the company’s largest planned training run remains on hold. OpenAI did not specify when the slowdown began or provide a timetable for resuming the work.
Unlock the Full Article
This article is exclusive to The Ledger Asia Subsribers / PAID members.
Already have an account? Log in here