OpenAI Reminds New High Risk Models For Cybersecurity, Prepares Layered Defense Systems

OpenAI warns that its latest generation of artificial intelligence models has the potential to pose a huge risk to cybersecurity (photo; x @_hatoya)

JAKARTA - OpenAI warns that its latest generation of artificial intelligence models has the potential to pose a big risk to cybersecurity, as AI capabilities increase rapidly and complex.

In a blog post on Wednesday, December 10, the Microsoft-backed company expressed concern that upcoming models could develop a zero-day exploit capable of penetrating high defense systems, or assisting operations to penetrate-level hacking to industries with real impacts on the physical world.

The warning comes amid global discussions on how AI can be used offensively by bad actors. OpenAI considers its ability to develop beyond the laboratory experiment scenario and is starting to approach "real-world capabilities" that could have serious consequences if it is misused.

OpenAI said it was investing in strengthening models for cyber defense tasks, including tools that make it easier for developers and security teams to audit code and patch vulnerability loopholes. Its mitigation strategy includes access control, infrastructure strengthening, data output restrictions, and stricter monitoring of model activities.

The company also announced plans to launch a new program that gives users or organizations that are considered eligible and focused on cyber defense. The access will include advanced capabilities that are considered too sensitive for general use.

To oversee AI frontier risk mitigation efforts, OpenAI will form the Frontier Risk Council, an advisory group filled with experienced cybersecurity practitioners. The board will work directly with OpenAI's internal team and will focus on cybersecurity risks before extending to other frontier capabilities domains.

Open warnings from OpenAI add to the urgency of discussions on how to maintain a balance between AI innovation and security, as this technology moves faster than the regulatory framework in many countries. Dialogues related to risk control, transparency, and AI governance are expected to come to light in the near future, as industries begin to face the dark side of generative artificial intelligence.