%201.webp)
Cowbell has launched a standalone AI liability insurance policy for US small and medium-sized businesses using AI agents to act on their behalf.
The policy is designed to cover both a business's own losses and third-party claims arising from what Cowbell defines as an "AI Incident". It will be sold through wholesale brokers on a non-admitted basis.
According to the insurer, the trigger includes cases in which an AI tool exceeds its authorised scope, acts without required approval, exposes data without authorisation, produces inaccurate or unsafe outputs, or fails to operate within established controls. It also covers adversarial manipulation, unstable behaviour, model drift and algorithmic discrimination.
The product targets a growing area of risk as businesses give AI systems more operational authority in functions such as customer service, payments and bookkeeping. The financial effects can range from misdirected payments and deleted records to legal claims from customers and other third parties.
Underwriting includes three AI-specific "Cowbell Factors" focused on observability, autonomy and governance. These are used alongside other underwriting information to assess how a business monitors its AI tools, what authority those tools have and what controls govern their use.
Coverage may include the cost of investigating and containing an incident, isolating the affected AI tool and preserving evidence. It may also cover correcting or withdrawing inaccurate or unsafe AI output, notifying recipients, and recovering from direct transaction losses linked to manipulation, unapproved actions or failed controls.
Additional areas include runaway usage costs caused by looping or escalating tasks, and "LLMjacking", in which unauthorised third parties use stolen credentials or API keys to generate AI usage charges. On the liability side, the policy may respond to defence costs, settlements and damages tied to claims alleging algorithmic discrimination, copyright or trademark infringement, bodily injury, or property damage linked to reliance on AI output or autonomous digital actions.
Coverage may also extend to regulatory proceedings and, where permitted by law, certain fines, penalties, consumer redress and punitive damages.
Claims scenarios
Cowbell outlined several examples of the incidents the policy is intended to address. In one scenario, an AI accounts payable agent misreads approval statuses and releases payments, including duplicate payments and payments for cancelled orders, without human sign-off. If a supplier later enters liquidation, part of those funds may not be recoverable.
Another example involves an AI quoting agent entering a loop and repeatedly launching additional agents while sending customers incorrect quotes. In that case, weak spending controls allow charges to mount before the business suspends the system and investigates.
A third scenario concerns an AI bookkeeping agent using outdated tax rates and filing client tax returns without human review. Clients then face penalties and interest, leading to reimbursement demands and lawsuits against the bookkeeping firm.
"SMEs are giving AI greater responsibility for how their businesses operate," said Trent Cooksley, Co-founder and Chief Operating Officer at Cowbell.
"The financial consequences can reach across the business, from money lost in a transaction to interrupted operations and customer claims. We designed Cowbell AI Liability Insurance to help businesses manage those risks as they put AI to work," Cooksley said.
The launch reflects a broader shift in insurance towards products that address operational risks from autonomous and semi-autonomous software, rather than only traditional cyber incidents such as data breaches or ransomware attacks. The distinction matters because AI tools can make decisions, initiate transactions and generate outputs that create losses even without a conventional network intrusion.
"AI liability is no longer only about people using AI. It's about AI agents acting on their own, with no human in the decision-making chain," said Simon Hughes, Chief Commercial Officer at Cowbell.
"When an agent goes rogue, a business can face its own losses and claims from customers and others at the same time. We designed Cowbell AI Liability Insurance for that agentic reality, giving brokers a concrete solution for their clients," Hughes said.
Businesses using tools such as Gemini, Claude and ChatGPT can access the AI-specific risk factors through Cowbell's connectors, according to the insurer.