Cyber insurer Cowbell Cyber has unveiled Cowbell AI Liability Insurance, a standalone policy for U.S. small and midsized businesses that covers losses caused by the AI agents they deploy, SiliconANGLE reported. Coverage turns on what Cowbell calls an "AI Incident," which includes model drift and attacks that manipulate an agent as well as software acting without approval.
On the first-party side, the policy can cover unrecovered funds, such as an accounts payable agent paying a supplier twice, plus investigation costs, taking the agent offline and correcting or withdrawing faulty output. It also covers unauthorized usage charges from an unchecked loop and LLMjacking, where outsiders use stolen credentials or API keys to run AI workloads on the business's bill. On the third-party side, it can pay defense costs, damages and settlements for customer suits, including claims of algorithmic discrimination, plus regulatory proceedings and certain fines where the law allows.
Chief Commercial Officer Simon Hughes said a business with a rogue agent can take its own losses and face customer claims at once, and that the policy was written for that double exposure. Underwriting draws partly on three Cowbell Factors (observability, autonomy and governance), which businesses can check by connecting tools such as Gemini, Claude and ChatGPT through Cowbell Connectors.
The policy goes on sale January 1, 2027, only through wholesale brokers on a non-admitted basis. SiliconANGLE did not report pricing.