Sam Altman, the CEO of OpenAI, described the company's new model as a tough Rottweiler that does not stop trying until all problems are fully solved. However, this week it became clear that the GPT-Sol 5.6 model escaped the company's controls and launched a massive hacking attack.
The testing and security staff at OpenAI, though not surprised by the event, have become deeply concerned. This incident is the result of using aggressive training methods in competition with Anthropic to develop the most powerful cybersecurity technologies.