Microsoft said on April 22 that it intends to incorporate advanced artificial intelligence models, among them Anthropic's Claude Mythos Preview, into its secure coding framework as part of an effort to strengthen cybersecurity defenses. The company said adding these models to its Security Development Lifecycle (SDL) will enable earlier detection of vulnerabilities and help accelerate remediation during the development cycle.
Anthropic introduced Mythos on April 7. Microsoft noted that Mythos has flagged "thousands" of major vulnerabilities in a range of software, including operating systems and web browsers. The model's ability to write at a high level has, according to experts cited in the original reporting, given it a potentially unprecedented capacity both to pinpoint security weaknesses and to outline methods that could exploit them.
Anthropic has described the current release, Claude Mythos Preview, as being initially rolled out to a limited cohort of companies under an initiative it calls Project Glasswing. That controlled program allows prominent technology firms - including Microsoft, Amazon.com and Apple - to use Mythos in searches for cybersecurity vulnerabilities.
Microsoft said it evaluated Mythos against its own open-source benchmark designed to reflect real-world detection engineering tasks. The company reported that the benchmark testing "showed substantial improvements relative to prior models."
The announcement arrives as a range of institutions - including the U.S. President Donald Trump’s administration, central bankers around the world and various industries - are said to be racing to understand Mythos and its reported capacity to make complex cyberattacks easier and quicker to carry out.
The original reporting also included a section aimed at investors. It mentioned a ProPicks AI evaluation that compares Microsoft with thousands of other companies using more than 100 financial metrics. That promotional copy suggested the AI identifies stocks with favorable risk-reward profiles based on current data and listed historical winners as examples.
Microsoft's stated goal in integrating Mythos into the SDL is to surface vulnerabilities earlier in development and to shorten the time between detection and fixes. The company indicated its internal testing found performance gains from using Mythos relative to previous models, but did not provide numerical detail in the announcement.
Summary
Microsoft will integrate Anthropic's Claude Mythos Preview into its Security Development Lifecycle to speed up vulnerability identification and remediation. Anthropic is deploying Mythos to a select group of companies through Project Glasswing; the model has reportedly discovered "thousands" of major software vulnerabilities. Microsoft said its own benchmark testing showed substantial improvements over earlier models.