News
Rubrik Taps Anthropic's Mythos Preview to Hunt Software Vulnerabilities
Rubrik is using Anthropic's Mythos Research Preview to defensively test its software as part of Project Glasswing, a security initiative focused on finding and fixing vulnerabilities in critical software infrastructure before they can be exploited.
The company announced Tuesday that it has gained access to Anthropic's Mythos Research Preview, describing the move as part of the "evolution of AI and cybersecurity." Rubrik said it is testing the advanced frontier AI defensively to identify, vet and patch potential software vulnerabilities across its enterprise platform and product suites.
The announcement ties Rubrik, which describes itself as the Security and AI Operations Company, to Anthropic's Project Glasswing, under which Anthropic is providing access to Mythos Research Preview to organizations that build or maintain critical software infrastructure.
Rubrik said the goal is to use the model before potential vulnerabilities can be used against the company or its customers. Bipul Sinha, CEO, chairman and co-founder of Rubrik, said in the announcement that Rubrik is putting Anthropic's model to work directly on its code and is "proactively purging potential vulnerabilities."
Defensive Testing With Frontier AI
Rubrik's announcement says Mythos Research Preview will be used to test the company's software defensively, not as a customer-facing product feature. The company said the model will be applied across Rubrik's enterprise platform and product suites for vulnerability identification, review and patching.
Rubrik expanded on that point, saying it has joined Project Glasswing and has access to Anthropic's Claude Mythos Preview at what it called a critical moment for organizations that build or maintain critical software infrastructure.
Anthropic's technical assessment describes Claude Mythos Preview as a new general-purpose language model that performs strongly overall and is "strikingly capable at computer security tasks." Anthropic said its testing found the model capable of identifying and exploiting zero-day vulnerabilities in every major operating system and every major web browser when directed by a user.
Anthropic said it launched Project Glasswing in response to those capabilities, describing it as an effort "to use Mythos Preview to help secure the world's most critical software, and to prepare the industry for the practices we all will need to adopt to keep ahead of cyberattackers." The company said the model is being released initially to a limited group of critical industry partners and open-source developers so defenders can begin securing important systems before models with similar capabilities become broadly available.
Recovery Framing
The company frames the announcement around preemptive recovery, saying Anthropic previously disclosed that a general-purpose frontier model had found thousands of critical software vulnerabilities across browsers, operating systems and other parts of the cyber landscape.
Rubrik said the Mythos Preview model shows both AI's ability to rapidly find vulnerabilities and the shrinking interval between vulnerability discovery and exploitation. The company summarized that point with the question: "How fast can you recover?"
Rubrik said the increased speed of vulnerability discovery reinforces its existing cyber-resilience positioning. The company said it built a unified architecture with preemptive recovery as the foundation, including immutable and air-gapped protection for data, identity and AI.
The company's message is that cyber resilience must move earlier in the process. In Rubrik's framing, recovery planning cannot start after an attack; it must be prepared before an incident occurs.
How Rubrik Says It Will Use Mythos Preview
Rubrik said Mythos Preview will help it rapidly find vulnerabilities in its platform, fortify its environment and reduce risk for customers. As part of the Project Glasswing cohort, Rubrik also said it will share findings back to the industry.
The announcement positions the work as an internal defensive testing effort and a broader statement about AI's role in cybersecurity. Rubrik said the combination of faster vulnerability discovery and AI-enabled attacks makes resilience and recovery speed a central security issue.
Rubrik's Broader AI Security Positioning
The announcement also connects to Rubrik's broader AI operations messaging. Rubrik said its Security Cloud secures, monitors and recovers data, identities and workloads across clouds, while Rubrik Agent Cloud is designed to accelerate trusted AI agent deployments by monitoring and auditing agentic actions, enforcing real-time guardrails, fine-tuning for accuracy and undoing agentic mistakes.
Rubrik's use of Mythos Research Preview adds a model-assisted vulnerability-discovery effort to that positioning. The company is presenting the work as a defensive application of frontier AI, focused on reducing software risk before attackers can take advantage of newly discovered vulnerabilities.
Rubrik plans to use the model to fortify its environment and reduce risk for customers. It also said the company is "ready for resilience."
About the Author
David Ramel is an editor and writer at Converge 360.