A few weeks ago, Anthropic introduced their latest language model called Claude Mythos that can find security vulnerabilities in various digital systems, including websites and software.
The system has been used by several parties to determine security vulnerability issues in their services, and some companies like Mozilla have used it to find various types of vulnerabilities in just a few seconds.
Recently, OpenAI has launched Daybreak, a digital systems and software security initiative that uses AI technology to scan their customers' digital systems for security vulnerabilities that allow them to improve the security of their software.
Daybreak combines the intelligence of OpenAI models and uses Codex as an agentic platform to allow users to see and determine various vulnerabilities that can be found in their systems.
In addition, Daybreak can also be used for secure code reviews, threat modeling, patch verification, dependency risk analysis, detection and remediation guidance into the daily development loop so that their software is safer to use.
OpenAI also said that they will work with the US government and their industry partners in the near future to determine if Daybreak can be used on a wider scale.
