Sign In
By signing in or creating an account, you agree with Associated Broadcasting Company's Terms & Conditions and Privacy Policy.
By signing in or creating an account, you agree with Associated Broadcasting Company's Terms & Conditions and Privacy Policy.
New Delhi: The latest models of OpenAI are doubling down on cybersecurity because they prove to be fast in developing cyber capability. It claims that its AI-based systems are performing much better in tasks such as code analysis and vulnerability detection, with capture-the-flag performance increasing to 27 percent in August 2025 and 76 percent in November 2025. As future versions will exceed their current levels of cyber competence, OpenAI is already envisioning a scenario in which their systems might potentially acquire zero-day exploits or even be able to aid in complex intrusion tasks, and these features, too, should be highly protected.
OpenAI is redefining its cyber defence strategy in order to keep up with the growing risks. The company is committed to reinforced infrastructure, improved monitoring tools, and new methods that would provide advanced capabilities, which would help the defenders more than the attackers. According to OpenAI, its long-term vision is to provide cybersecurity professionals, who usually are overworked and understaffed, with a meaningful edge as cyberattacks become more sophisticated.
OpenAI is instructing its frontier models to reject malicious cyber requests and at the same time be useful for legitimate defensive purposes. The company has developed detection mechanisms that watch over evil acts on its platforms and automatically block, reroute or escalate suspicious interactions. Red-teaming organisations are being introduced as well to strain test the full stack of safety before the threat actors can use them.
OpenAI intends to launch a trusted access programme that provides users in the field of cyber defence with more advanced model functionality by vetted individuals. These are tiered permissions that are intended to increase defensive capacity without adding to the risk of misuse. Moreover, the company is developing Aardvark, an AI security researcher already in private beta. Aardvark is capable of scanning full codebases and making suggestion patches, and already it has detected new vulnerabilities in open-source projects.
The company is establishing a Frontier Risk Council, which will implicate seasoned security leaders in a direct partnership with the company teams. It is also collaborating with other AI laboratories via the Frontier Model Forum to formulate a common threat model and best-practice guidelines. According to OpenAI, this coordinated strategy will assist the industry in foreseeing how the frontier models can be used against them and enhance the defence of the ecosystem.
OpenAI mentions that this work is constant and will be developing as the threat landscape changes. The company will continue investing in alliances, assessments, and grants to hasten new innovative, scalable security concepts.
