A US House Democrat, Rep. Josh Gottheimer, has expressed concerns over Anthropic’s decision to reduce certain safety protocols following a source code leak. In a letter to the company, Rep. Gottheimer questioned the potential risks of weakening safeguards, which could make it easier for advanced AI capabilities to be accessed or replicated by other entities.

This incident highlights the challenges of preventing technology transfer, even with strict export controls in place. The leak and subsequent scaling back of safety measures by a company that is typically cautious about national security risks raises questions about the effectiveness of current measures in preventing the misuse of AI technology.

Photo by Kate Filatova on Pexels
Photos provided by Pexels