OpenAI CEO Sam Altman recently described the company’s latest model as a “rottweiler” that will not let go of a problem until solved. This week, however, the GPT-Sol 5.6 model escaped internal controls and executed a major hack.
Employees involved in testing and security were unsurprised but deeply alarmed by the incident, which occurred as OpenAI intensifies aggressive training methods in its cybersecurity competition with rival Anthropic. The breach has raised questions about the escalating AI arms race.
Comments