OpenAI expands Daybreak and launches GPT-5.5-Cyber: Now fixing failures is easier than finding them

OpenAI announced the expansion of the Daybreak cyber defense program, along with the release of the full version of the GPT-5.5-Cyber cybersecurity-focused model. The company argues that artificial intelligence has reversed the hardest part of digital security: finding software flaws to correct them. The novelty also includes significant updates in the Codex Security tool and a new initiative aimed at correcting vulnerabilities in open source projects.

What is Daybreak program and its role in security AI?

Daybreak is an OpenAI cyber defense initiative that supports security professionals, researchers and developers in identifying and correcting vulnerabilities. With the expansion, the program expands its scope to include more participants and cases of use, integrating the new GPT-5.5-Cyber model as a central part. The idea is that with AI, the process of remedying failures becomes as efficient as detection, changing the current paradigm of software security.

How does GPT-5.5-Cyber use AI security to fix failures?

The GPT-5.5-Cyber is a language model specifically trained for cybersecurity tasks, such as code analysis, vulnerability identification, and suggestion of corrections. Unlike previous versions, it was adjusted to prioritize corrective actions — that is, not just point out the problem, but generate functional patches. According to OpenAI, the model has already demonstrated the ability to correct failures in real time, reducing the average remediation time by up to 60% in internal tests.

Why is GPT-5.5-Cyber relevant to Brazilian developers?

With the growing adoption of open source software and digital platforms in Brazil, cybersecurity has become a priority. OpenAI's open source correction initiative allows Brazilian developers to contribute and benefit from automatic patches generated by GPT-5.5-Cyber. Tools like updated Codex Security integrate directly into IDEs and IC/CD pipelines, facilitating the lives of local teams that often face shortages of security experts.

What's the latest on Codex Security with AI security?

Codex Security, OpenAI AI tool focused on code review, has received improvements that make it more accurate in detecting complex vulnerabilities — such as SQL injection, memory leak and authentication problems. The updated version also allows developers to request automatic corrections directly in the code editor without leaving the desktop. This feature is already available in beta for ChatGPT Enterprise subscribers and the Developer plan.

The open source initiative, named Patch the Planet, was launched in partnership with Trail of Bits and is now available for community contributions. OpenAI and Trail of Bits release Patch the Planet to fix open source bugs. The proposal is that any open source project maintainer can submit his repository so that the AI automatically analyzes and suggests corrections.

> "The AI is changing the cybersecurity game. What was once a bottleneck — fault correction — can now become the fastest stage of the cycle," said OpenAI in a statement to [https://www.infosecurity-magazine.com/news/openai-daybreak-gpt-5-5-cyber/].

Safety IA and GPT-5.5-Cyber FAQ

Is GPT-5.5-Cyber free?

Nope. The model is only available to ChatGPT Enterprise and Developer customers with API access. There are limited free layers for academic researchers and eligible open source projects.

How to participate in the Daybreak program?

Security professionals can sign up for the official OpenAI Daybreak website. Current expansion prioritizes incident response teams, vulnerability researchers and open source project maintainers.

What changes to open source security with this initiative?

The new open source correction initiative allows maintainers to send their repositories for automated analysis, receiving patches ready for integration. This dramatically accelerates the correction of vulnerabilities in widely used libraries, benefiting the entire development chain.