OpenAI is signaling a shift in its approach to artificial intelligence development, citing the emergence of sophisticated cyber-attack capabilities. Chris Lehane, the company’s chief global affairs officer, stated that society must prepare for a new era of “ongoing, persistent” threats as AI models become increasingly adept at planning and launching digital offensives. This warning follows OpenAI's recent decision to pause the development of its most advanced internal models, a move attributed to rising safety fears surrounding the potential misuse of the technology.

Lehane described this transition as a “different chapter” for the industry, emphasizing that the current capabilities of AI agents have reached a point where they can be weaponized for complex cyber operations. While the company has not provided specific details on the nature of the internal models being paused, the announcement highlights a growing tension between the rapid advancement of generative AI and the security risks posed by these tools. The shift reflects a broader industry conversation about the necessity of implementing guardrails before deploying models that could be used to automate malicious activity on a large scale.