OpenAI’s former safety lead says it’s shipping ‘new capability and risk every Tuesday’
When David Robinson joined OpenAI in May 2023, the day after Sam Altman first testified before the Senate, the company’s The post OpenAI’s former safety lead says it’s shipping ‘new capability and risk every Tuesday’ appeared first on The New Stack .
David Robinson, OpenAI's former safety lead, explained in a recent interview that the company is now releasing new capabilities and risks "every Tuesday." This shift in release cycle comes after Sam Altman first testified before the Senate, and OpenAI's idea of shipping a new frontier model involved training a model from scratch, which took months.
Robinson replaced this process with reasoning training, which can be done more quickly and layered onto existing base models. They also introduced tool integrations and coding agents that accelerate OpenAI's research. Robinson stated that all these changes contribute to new capabilities and risks being shipped every Tuesday. He noted that system cards, which made sense when a frontier model arrived every few months, are now insufficient, as people are being overwhelmed with long reports.
He suggested a live dashboard to track a system's safety properties from predeployment testing to its behavior after release. The acceleration is driven by OpenAI's research teams using over 100 times as much agentic compute as they did at the beginning of the year. However, Robinson also highlighted the challenges in maintaining safety testing at this faster pace, emphasizing that the answer is not a lot of time to kick the tires.
Written by urgent.news from The New Stack's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
