OpenAI has issued a warning regarding the potential cybersecurity risks associated with the next generation of its AI models, which could perform tasks previously reserved for advanced hacking groups. In response to competitive pressures from Google’s Gemini project, OpenAI emphasizes a commitment to ensuring that its technology is used defensively, investing in capabilities that assist cybersecurity teams with tasks like auditing and vulnerability patching. The company plans to implement stricter access controls and establish the Frontier Risk Council to collaborate with cybersecurity experts. However, internal reports indicate an urgency to accelerate development through increased reliance on real-time user feedback, which could lead to inconsistencies in training data, presenting both risks and opportunities for rapid advancement in AI capabilities.