OpenAI’s Daybreak harnesses the smarts of AI to find security flaws

Applications of AI


OpenAI announced Daybreak, a suite of cyber defense tools that uses the intelligence of the company’s LLM to find and remediate software vulnerabilities.

The company hasn’t released the tool freely to everyone. Instead, it will “collaborate with industry and government partners” as it prepares to deploy an “increasingly cyber-enabled model” in the future.

Mashable 101 Fan Favorites: Vote for your favorite creators today.

According to OpenAI, Daybreak combines the intelligence of OpenAI models (particularly the new GPT-5.5) with the extensibility of Codex, OpenAI’s coding tool. Partners can use Daybreak for “secure code reviews, threat modeling, patch validation, dependency risk analysis, detection, and remediation guidance.”

Three models are currently offered. The default GPT-5.5 model can be used for general purpose work. GPT-5.5 with Trusted Access For Cyber ​​is for “the most defensive security workflows” including secure code review, malware analysis, and patch validation. Finally, GPT-5.5-Cyber ​​can be used for authorized red teaming, penetration testing, and controlled verification.

See also:

OpenAI deploys ChatGPT 5.5 Instant as new default model for everyone

If this sounds familiar, you’re probably thinking of Anthropic’s Project Glasswing initiative and Mythos, an AI that’s very good at finding vulnerabilities in software that Anthropic allows only select partners to access.

OpenAI has not disclosed pricing for Daybreak. If you are interested, please contact the company’s sales team for a quote. Several companies are listed as partners, including Cloudflare, Cisco, Oracle, and Akamai.

Topics
Artificial Intelligence OpenAI



Source link