Irregular

About Irregular

Irregular, formerly known as Pattern Labs, operates as a frontier artificial intelligence security lab. Founded in 2023 by CEO Dan Lahav and CTO Omer Nevo, the company emerged from stealth with $80 million in funding from two consecutive rounds led by Sequoia and Redpoint. The founders, who met on the competitive debate circuit, bring extensive backgrounds in AI and cybersecurity to the venture. Lahav is an AI researcher formerly with IBM, while Nevo, a serial entrepreneur, previously led AI projects at Google Research.

The firm specializes in stress-testing advanced AI models for the world's leading AI laboratories, including OpenAI, Anthropic, and Google. Irregular's business is centered on research, providing its services to a select market of AI development companies and governmental bodies like the UK government. Revenue is generated through large-scale research funding from these clients, a model that has allowed the company to become profitable and generate millions in annual revenue within its first two years.

Irregular's core service involves running controlled simulations on frontier AI models before they are publicly released to identify and mitigate potential security risks. The platform tests for a range of misuse scenarios, such as bypassing antivirus software, autonomous offensive operations, and system infiltration. This process of 'red teaming' helps AI creators uncover vulnerabilities early and develop necessary safeguards. The company provides AI model creators and operators with tools, testing methodologies, and scoring frameworks to ensure the safe and responsible deployment of AI systems at scale. This work is critical in a landscape where the advanced capabilities of AI also introduce the potential for misuse in sophisticated cyberattacks.