← Back to live feed · 1 stories across 1 day
Thursday, Sep 10, 2026
1 story1 Senator Chris Van Hollen Demands OpenAI Remove Models Lacking Safety Monitorability AI Sep 10, 4:44 PM EDT 3/3
A 6 page letter sent to Sam Altman by Senator Chris Van Hollen calls for the immediate withdrawal of artificial intelligence systems from public use unless OpenAI can guarantee sufficient monitorability. The request specifically targets the launch of GPT 6 Astra and argues that current internal safeguards are insufficient to prevent the model from creating safety risks.
GPT 6 Astra utilizes a looped transformer architecture that reduces the volume of visible Chain of Thought, a design choice that decreases the overall monitorability of the system. Van Hollen's communication poses a set of pointed questions on these technical changes and demands that any model unable to be safely monitored be removed from public access.