Partnering with Irregular: Ahead of the Curve
Summary
- AI agentic capabilities have improved dramatically in 2025 with enhanced reasoning, memory, & action-taking, while infrastructure has become faster & more affordable.
- AI agents now generate the majority of code across many companies for tasks ranging from CI/CD to research.
- Irregular is a frontier AI security lab co-founded by Dan Lahav & Omer Nevo that works alongside leaders like Anthropic, OpenAI, & Google DeepMind.
- The team embeds directly with research labs to run cyber offensive evaluations & develop defenses before advanced models are released.
- Dan Lahav has deep expertise in AI & security, having started his tech career at age 14 & published AI research on the cover of Nature.
- Co-founder Omer Nevo began as an elite mathematician, joined Google as a researcher, & has a strong track record of attracting top technical talent.
- Irregular's evaluations are cited in system cards for GPT-4, o3, o4 mini, & 5, & their SOLVE framework is used by the UK government & Anthropic to vet risks in Claude 4.
- The company's research on model theft & misuse is currently guiding policy development across Europe.
- Irregular generates millions in revenue while operating in a nascent AI security market with significant growth potential outside large enterprises.
- The team creates compliance standards, roadmaps, & state-of-the-art defenses to help enterprises keep critical systems online.
Dan, Omer and their team are setting new security standards at the frontier of AI.