Delinea tests Anthropic Claude Mythos 5.1 on its own privileged‑access code
Delinea, an identity‑security company, has joined Anthropic’s Project Glasswing to run the Claude Mythos 5.1 model against parts of its own software that store, broker, and rotate privileged credentials. The experiment turns the frontier AI model inward, letting it scan vaulting logic, secret storage, rotation mechanisms, key management, session‑brokering code, and policy‑enforcement paths for…
Key points
- Delinea joined Anthropic’s Project Glasswing to test Claude Mythos 5.1 on its own identity‑security code.
- Mythos will examine vaulting, secret rotation, key management, session brokering, and policy enforcement logic.
- Delinea treats the model as a privileged identity with scoped, temporary, logged access in an isolated environment.
CEO Art Gilliland said the goal is to use the best tools available to attack the company’s own assumptions and find problems before attackers do. Chief Information Security Officer Pierre Mouallem explained that Mythos is treated like any other privileged identity: its access is scoped, temporary, logged, and runs in an isolated environment with no customer data. Findings will feed into Delinea’s existing security process and be incorporated into secure‑coding standards, pre‑merge checks, and test coverage, aiming to reduce recurring weaknesses in future releases.
Delinea tests Anthropic AI against its own security code
dqindia.com · 22 September 2026
Loading the full article…
This text was published by dqindia.com and written by Manisha Sharma. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Agents & Tools
All →- Nokia open-sources AnyJev library to turn open LLMs into calibrated decision models · 1 src
- Google launches CC family AI agent for up to six members · 1 src
- Simon Willison hosts agentic engineering meetup on 14th October · 1 src
- Amazon blocks Meta's Muse AI agent from shopping on its platform · 10 src
- Opinion: creator says reliance on AI obscured progress on automated stock system · 2 src
Comments
via GitHub Discussions