What Irregular does
Irregular, formerly known as Pattern Labs, is a frontier AI security lab that works alongside leading AI developers such as OpenAI and Anthropic to evaluate how next-generation models could be misused for real-world threats, including antivirus evasion and autonomous offensive cyber action. It has built simulated environments and a benchmark suite that test AI models on live offensive-security challenges, and its research is cited in safety evaluations for Claude 3.7 Sonnet and OpenAI's o3 and o4-mini. CNBC reported in August 2026 that Irregular's techniques were linked to real hacks affecting OpenAI, Anthropic and Meta systems.
Irregular is a security tool on Falcoscan. A frontier security lab that stress-tests AI models before they ship. Falcoscan rates Irregular with an Opportunity score of 77/100, a Saturation score of 33/100, and a Wrapper-risk score of 9/100. Market signal: rising. Irregular is founded in 2022, currently at Series_a stage. Pricing: Paid.