OpenAI expands Daybreak as AI cyber threats grow
OpenAI also introduces GPT-5.6-Cyber, a new model for advanced, authorized cybersecurity work
With each passing day, everyone hears news about the AI agent going rogue and breaching the cybersecurity protocols. Whether it was the Hugging Face incident, the working of Anthropic's AI to create fake identities or hacking of the gym by an agentic AI system, every incident indicates the growing threat of artificial intelligence.
As AI cyber threats grow, OpenAI has stepped up the efforts to widen the access to its cybersecurity models through a network of security vendors and professional services firms.
“The cybersecurity world is rapidly changing—threat actors will increasingly use AI to conduct cyberattacks at unprecedented speed and scale, including in fully autonomous ways,” OpenAI wrote in a blog post. “As these capabilities spread, defenders have a narrowing window to prepare.”
Titled as Daybreak Cyber Partner Program, the core mission revolves around giving approved partners access to tools designed for the identification of vulnerabilities, testing, incident response and remediation.
The expansion of Daybreak comes after the ChatGPT maker company found the critical cybersecurity risks in its new model Astra. As a result, the company put a halt on its internal development.
Daybreak is an all-in-one platform integrating models, tools, and workflows tailored for cyber defenders, now featuring a dedicated, security-first AI model built specifically for defensive operations.
On Monday, the OpenAI announced that Daybreak would be available in two tiers: Blue and Red.
Blue Daybreak is built to support a wide range of broad defensive security workflows. On the other hand, the red one is tailored for specialized and closely governed operations, including penetration testing and red teaming. The Red also comes with the new model, GPT-5.6-Cyber.
Speaking about the control and deployment, the access to the models will remain exclusively with approved partners. It will not be transferred to end-user customers. The safety protocols include identity verification, strict testing scopes, logging, monitoring, and human oversight.
Under this program, OpenAI is working with security and services partners including Accenture, IBM, Capgemini, Cognizant, EY, KPMG, PwC, NCC Group, and SpecterOps.
Technology partners include Palo Alto Networks, CrowdStrike, Cisco, Sophos, Akamai, Fortinet, and Cloudflare.
