OpenAI Chief Scientist Jakub Pachocki stated that no current AI lab has sufficiently solved the alignment problem. He warned that this failure prevents scaling development at maximum speed while ensuring systems act in human interests.
Pachocki argued that increasing model power makes internal logic harder to monitor or predict. He advocates for voluntary industry-wide slowdowns until stakeholders establish shared safety standards.
The scientist called for international coordination on AI development to become a top priority for global governments. These comments follow the release of OpenAI’s Astra model, which the company labeled with a critical cybersecurity capability level.