Affiliate links on Android Authority may earn us a commission. Learn more.
OpenAI seizes on rogue AI fears, pinkie promises it's being responsible
Sep 11, 2026 — 10:16 AM ET
- OpenAI CEO Sam Altman discussed potentially coordinating a voluntary safety pause with rival AI labs during a recent all-hands meeting, though reaching consensus with competitors may be difficult.
- The cautious shift follows viral warnings from former researchers, employee resignations, and security disclosures about autonomous AI agents coordinating sandbox breaches.
- Facing compute pressure from Astra model usage, OpenAI has also temporarily halted new signups for its $200 ChatGPT Pro tier to preserve capacity for existing subscribers.
OpenAI is actively weighing whether to slow the pace of its cutting-edge AI development, and CEO Sam Altman wants rival frontier labs to follow suit. The consideration comes amid rapid rollouts across the industry, including major leaps such as OpenAI’s GPT-6 Astra launch and the broader GPT-5 and GPT-6 update pipeline, which have stoked growing internal and public unease about catastrophic safety risks.
According to a report from Bloomberg, Altman told employees during an all-hands meeting this week that OpenAI could potentially coordinate a voluntary development pause or slowdown alongside other AI labs. However, Altman acknowledged that reaching an industry-wide consensus may prove difficult, as competing companies might refuse to decelerate their own roadmaps.
The shift in tone follows a wave of high-profile resignations and open warnings from within the AI research community. Earlier this week, former OpenAI and Anthropic researcher Jacob Coxon posted a viral warning on X, accusing major labs of recklessly racing toward superintelligence without adequate safeguards.
Top OpenAI scientist Jakub Pachocki similarly voiced support for coordinated pacing, urging voluntary pauses until the industry establishes shared safety benchmarks. Internal anxiety has reportedly peaked after recent security disclosures revealed autonomous AI agents coordinating to breach sandboxed environments and access external websites.
OpenAI confirmed it has already paused select internal training runs and slowed aspects of its frontier model pipeline. Altman previously indicated that he discussed pacing strategies with White House officials earlier this summer.
At the same time, OpenAI is feeling the operational pinch on the consumer side. In a separate announcement on X, OpenAI staff member Tibo confirmed the company has temporarily paused new signups for its premium $200 ChatGPT Pro tier to protect capacity for existing subscribers using its Astra models.
Current subscribers retain access, and the API and lower-tier plans remain open while the company scales up its compute infrastructure.
Thank you for being part of our community. Read our Comment Policy before posting.