OpenAI Faces Copyright and AI Safety Scrutiny Over Trump Administration Brief and Astra’s Reasoning Technique
aiinsider ·
OpenAI is facing two separate controversies involving copyright policy and AI safety. The Trump administration filed a brief supporting OpenAI in a copyright lawsuit brought by The New York Times, arguing that American AI leadership is a national interest and warning against misapplying fair use doctrine. The filing is not a ruling, and the case is still before a federal court, though prior copyright cases have generally favored AI companies. Separately, reports indicate OpenAI's upcoming Astra model uses a technique called 'recurrent depth' that processes queries in loops, potentially making its reasoning harder to monitor. Safety researchers, including those from Redwood Research and commentator Zvi Mowshowitz, expressed concerns that this could undermine chain-of-thought transparency across the industry. OpenAI's chief scientist stated the company remains committed to preserving legible reasoning traces, and the technique's current use is reportedly limited. However, Anthropic and Google DeepMind are also reportedly exploring similar approaches, raising broader questions about the future monitorability of AI reasoning.
aiinsider