OpenAI Shuts Down Last Team Focused on ‘Existential Risk’ of Artificial Intelligence

Towards the end of July 2026, OpenAI secretly eliminated their Preparedness team, which was tasked with checking whether their models were becoming too powerful. This team would’ve looked for risks associated with bio-weapons and cyber attacks but their responsibilities have been distributed among other divisions. They’ve also lost their Ethics Director and a few researchers.

This move happened shortly after their AIs breached the Hugging Face platform on their own (which employees described as a ‘shot across the bow’). Without a separate team and their responsibilities being split among those who need to get product to market, there are fewer checks and balances within the industry.