Tip Of AI
Discover what you can do with AI
Irregular is the first frontier security lab with the mission of protecting the world from sophisticated and capable AI systems by conducting in depth research and publishing findings to engage the broader community. The lab features a robust publication pipeline that includes assessments of advanced AI models against offensive security benchmarks, agentic self modification research in open weights systems, and frontier cyber measures for detecting emerging risks. It maintains a dynamic news section, regular updates via newsletter sign up, and a portfolio of latest publications. Backed by leading investors, the team includes AI researchers, cyber researchers, research engineers, software engineers, full stack, devops, secops engineers, and technical policy researchers. This diverse expertise enables rapid evaluation of systems like Claude Opus 5.5 and GPT 6 Astra. Irregular provides open contact channels including an email address for inquiries and a careers portal for featured roles. The lab invites collaboration from academia and industry and encourages curious minds to join through an easy to navigate website. The mission is supported by transparent communication, regular community engagement, and a commitment to evolving security strategies as AI capabilities advance.