Armilla AI Expands Coverage for AI Risk to $25M
Armilla AI, backed by Lloyd’s, has increased its coverage for AI model failure and liability risks to $25 million. The move comes as traditional insurers reportedly retreat from underwriting what they consider "unbounded" AI risk. This signals growing demand for insurance products specifically tailored to algorithmic systems and their unique operational and legal exposures.
- Armilla AI's coverage is designed to address specific AI failures not typically covered by traditional insurance, such as malfunctions, "hallucinations," critical inaccuracies, or a general failure to perform as intended. This new AI Liability Insurance policy is underwritten by Lloyd's, including partners like Chaucer and Axis Capital. - The expanded $25 million limit is part of a product called Vanguard AI, a collaboration between Armilla AI and Chaucer Group. This offering combines Chaucer's cyber and technology errors and omissions coverage with Armilla's distinct AI liability policy. - The policy covers legal fees, settlements, and third-party liabilities that arise when an AI system's underperformance leads to operational disruptions or reputational damage. It also now extends to AI-driven property damage, regulatory violations under acts like the EU AI Act, and data leakage that doesn't involve a breach. - This specialized AI insurance aims to fill a "coverage vacuum" being created as traditional insurers add exclusions for generative AI exposures to their general liability, cyber, and errors and omissions policies. Major carriers like AIG and W. R. Berkley have begun to introduce such exclusions due to the systemic and hard-to-quantify risks of AI. - Before issuing a policy, Armilla's platform first assesses the AI model for risks like racial or gender bias, security vulnerabilities, and potential for hallucinations, leveraging global standards such as the EU AI Act and the NIST AI Risk Management Framework. This is based on data from over 500 AI evaluations across various regulated industries. - In addition to insurance for enterprises using AI, Armilla also offers a warranty product for AI developers called "Armilla Guaranteed." This allows developers to assure customers that their AI solution will meet specific performance metrics, with a financial remedy if it fails. - Toronto-based Armilla AI was founded in 2020 and has raised a total of $6.37 million in funding from investors including Y Combinator, Mistral Venture Partners, and MS&AD Ventures. The company is a graduate of the Lloyd's Lab accelerator program.