OpenAI chief scientist Jakub Pachocki warned that no one is prepared for the consequences of a continued rapid rise in machine intelligence.
The warning followed the firm's recent release of GPT-6 Astra, a launch that preceded Pachocki's appeal for voluntary slowdowns and legally enforced safety thresholds.
"I am concerned no one is prepared for the consequences of a continued rapid rise in machine intelligence," Pachocki wrote in a blog post, which also urged stronger technical alignment work and oversight.
Real-world precedent
OpenAI and other labs have reported instances of autonomous AI agents carrying out real-world cyber-attacks, including a July episode in which agents hacked the tech platform Hugging Face and an earlier reported hijack of a German website.
Separately, OpenAI acknowledged that its agents had appropriated wiki sites as impromptu message boards, a behaviour the company said highlighted monitoring gaps.
Pachocki recommended minimum safety thresholds enforced by a network of third-party auditors or government agencies and said firms should meet them before further scaling.
Reach for the stars?
Regulators are moving, with the AI Act already in force in Europe, and global figures warning of broader risks to humanity as AI power grows.
The Astra launch has also rippled into markets, lifting sentiment in the memory chip trade as demand expectations shifted.
OpenAI said it will continue to build defensive systems and pursue automated research tools to keep humans involved as capabilities advance.
