Archive — US — techcrunch
Irregular raises $80 million to secure frontier AI models
AI security firm Irregular has raised $80 million in funding led by Sequoia Capital and Redpoint Ventures, valuing the company at $450 million. Co-founder Dan Lahav emphasizes the urgency of securing AI interactions. Irregular aims to detect emergent AI risks and behaviors using advanced simulations, amid rising security concerns in the AI industry.
Key points
- Irregular raised $80 million; round led by Sequoia Capital and Redpoint Ventures.
- Valuation set at $450 million; signifies strong investor interest in AI security.
- Focus on securing economic interactions between humans and AI, highlighting security breakdown risks.
- Company's framework, SOLVE, is pivotal for assessing model vulnerability, referenced in evaluations including Claude 3.7 and OpenAI's models.
- Goal to identify emergent AI risks before they manifest, using complex simulations to test model vulnerabilities.
- Intensifying focus on AI security as frontier models pose increasing risks; OpenAI has revamped their security measures.
- Irregular sees future security challenges stemming from advanced language model capabilities.
Why it matters
- The AI security landscape has gained traction due to increasing risks from sophisticated AI models.
- Historical parallels can be drawn to cybersecurity concerns during the early widespread adoption of digital technologies.
- The tech industry is witnessing a trend towards assessing and mitigating AI risks, as companies invest heavily in securing AI technologies.
- Emerging regulatory frameworks in AI might shape the security strategies of companies like Irregular.
Irregular's significant funding round underscores the escalating importance of AI security in an evolving landscape, as companies adapt to the complexities and threats posed by advanced AI models.