Insurers are adding fresh restrictions to payouts for cyber incidents and regulatory penalties linked to AI, reflecting efforts to curb potential liabilities tied to the technology’s development.
QBE and Beazley are among the insurers that have drafted policy wording intended to place tighter limits on certain AI-related claims, reported the Financial Times (FT), citing brokers and documents.
Go deeper with GlobalData
Access deeper industry intelligence
Experience unmatched clarity with a single platform that combines unique data, AI, and human expertise.
QBE has introduced “sub-limits” for large language model (LLM) jacking incidents, a form of abuse in which criminals break into LLMs used by companies to sidestep usage charges.
Such sub-limits set a lower ceiling for a particular category of loss.
Under QBE’s wording, a cyber policy with total cover of up to $5m would provide only around $250,000 for losses arising from LLMjacking.
Beazley has also put forward contract terms designed to restrict its liability for AI-related losses associated with regulatory breaches.
Insurers maintain that these sub-limits confirm AI risks are included within policy cover.
However, brokers and lawyers advising corporate policyholders said they were uneasy that the provisions could later be applied more broadly, reducing protection against other developing AI-related risks beyond LLMjacking.
Brokers and the documents seen by the FT indicate that, for both insurers, the AI sub-limits would hold payouts for specified AI losses to roughly 10% of the overall policy value.
Beazley said the policy amendments reviewed by the FT were still being developed and had not been added to any active policies.
In a statement to the FT, QBE said it was “not retreating” from AI risks.
“Where coverage developments have occurred, they have been focused on enhancing protection for specific emerging exposures… rather than narrowing core cover,” it added.
“If an AI-related event leads to a conventional cyber incident, resulting losses continue to fall fully within the core cyber policy.”
Separately, insurers including New York-based AIG have previously sought approval from US regulators to exclude AI-related losses from corporate insurance policies, as per FT reports.
