Irregular Secures $80M to Enhance Security of Advanced AI Models

Irregular Raises $80 Million to Bolster Security for Advanced AI Models

AI security firm Irregular announced on Wednesday that it has secured $80 million in a new funding round led by Sequoia Capital and Redpoint Ventures, with additional participation from Wiz CEO Assaf Rappaport. Sources familiar with the deal estimate the company’s valuation at around $450 million.

Irregular, formerly known as Pattern Labs, has established itself as a key player in AI model security and evaluation. Its proprietary framework, SOLVE, which assesses a model’s ability to detect vulnerabilities, is widely adopted across the industry. The company’s evaluations have informed security assessments for notable models such as Claude 3.7 Sonnet and OpenAI’s o3 and o4-mini.

Co-founder Dan Lahav emphasized the evolving nature of AI security challenges, stating, “Our view is that soon, a lot of economic activity is going to come from human-on-AI interaction and AI-on-AI interaction, and that’s going to break the security stack along multiple points.”

Proactive Identification of Emerging Risks

With this funding, Irregular aims to advance beyond identifying known vulnerabilities to detecting emergent risks and behaviors before they appear in real-world applications. The company has developed sophisticated simulated environments where AI systems engage in attacker-defender scenarios, enabling rigorous pre-release testing of new models.

Co-founder Omer Nevo explained, “We have complex network simulations where we have AI both taking the role of attacker and defender. So when a new model comes out, we can see where the defenses hold up and where they don’t.”

The focus on AI security has intensified industry-wide, especially as frontier AI models demonstrate enhanced capabilities that could be exploited. OpenAI recently revamped its internal security infrastructure in response to concerns such as corporate espionage. Simultaneously, AI’s growing proficiency in uncovering software vulnerabilities underscores the critical importance of robust defensive measures.

Ongoing Challenges in Securing Frontier AI

Irregular’s leadership acknowledges that securing advanced AI models is an ongoing and escalating challenge. Lahav remarked, “If the goal of the frontier lab is to create increasingly more sophisticated and capable models, our goal is to secure these models. But it’s a moving target, so inherently there’s much, much, much more work to do in the future.”