OpenAI Chief Scientist Warns of Unchecked AI Acceleration

Jakub Pachocki, OpenAI’s chief scientist, posted that the speed of artificial-intelligence progress is outpacing human comprehension and control. He cautioned that models are already capable of operating computers, collaborating with other AIs, and conducting research projects, and that “recursive self-improvement” could soon allow systems to upgrade themselves without human input. Pachocki urged “extreme caution” and advocated for voluntary slowdowns until shared safety standards are in place.

Bloomberg

Despite the warning, investors are pouring capital into self-improving AI ventures. Inherent, a startup founded by former DeepMind researchers, secured $50 million earlier this year. Recursive Superintelligence, led by former Meta scientist Richard Socher, attracted $650 million. OpenAI itself has announced a goal to create a “true automated AI researcher” within two years, underscoring the sector’s rapid push toward more autonomous systems.

Bloomberg

Other AI leaders have echoed concerns about societal impact. Anthropic CEO Dario Amodei warned that AI could replace half of all entry-level jobs, while OpenAI CEO Sam Altman recently told Bloomberg TV that the industry has not communicated enough about AI’s potential benefits. In contrast, Nvidia CEO Jensen Huang proclaimed on X that “AGI has arrived,” referencing OpenAI’s latest GPT-6 Astra model, which the company released only in a limited fashion due to its advanced cybersecurity capabilities.

Bloomberg

• “This is a time that calls for extreme caution,” — Jakub Pachocki, chief scientist

Bloomberg