Irregular raises $80 million to secure frontier AI models

AI security firm Irregular has secured $80 million in new funding, led by Sequoia Capital and Redpoint Ventures, at a reported $450 million valuation. This capital injection will enable the company to expand its focus from existing AI model vulnerability detection, where its SOLVE framework is widely used for models like Claude and OpenAI, to proactively identifying emergent risks through elaborate simulated environments. The investment underscores the escalating strategic importance of robust AI security as frontier models advance and the industry grapples with evolving threats.

Analysis

AI security firm Irregular has secured $80 million in a new funding round led by premier venture capital firms Sequoia Capital and Redpoint Ventures, achieving a reported valuation of $450 million. This investment validates the growing importance of the AI security sub-sector. Irregular is already an established entity in AI model evaluation, with its proprietary SOLVE framework being utilized in the security assessments for leading models such as Anthropic's Claude 3.7 Sonnet and OpenAI's o3 and o4-mini. The infusion of capital is designated to pivot the company's focus from evaluating existing risks to proactively identifying 'emergent risks' in frontier AI models before they are publicly deployed. This will be achieved through a sophisticated system of simulated environments where AI agents serve as both attackers and defenders. The transaction occurs amidst heightened industry focus on AI security, as the increasing capability of models to find software vulnerabilities presents a dual-use dilemma for both offensive and defensive cybersecurity operations.

Key Decisions for Investors

• The successful funding round led by top-tier VCs like Sequoia validates the 'picks and shovels' investment thesis within the AI sector, suggesting investors should increase monitoring of publicly-traded cybersecurity firms that are developing specific AI security solutions.
• Investors with exposure to large-cap AI developers (e.g., Microsoft, Google, Amazon via their cloud and AI partnerships) should view the growth of third-party validation firms like Irregular as a critical de-risking factor for the ecosystem, as robust security protocols can mitigate long-tail risks associated with deploying frontier models.