TL;DR

Specialist insurer CFC has added affirmative AI wording to its media policy from 30 July, completing a programme begun in June that now covers seven products including technology errors and omissions, professional liability, eHealth, intellectual property and management liability. Under the revised wording, AI involvement does not prevent the policy responding — so defamation, intellectual property infringement and other media and professional liability claims remain covered where AI contributed to the content.

The cyber wording was updated too, clarifying how established triggers apply when AI hallucinations or model prompts cause unexpected systems downtime.

“AI is no longer an emerging technology for media companies. It is already embedded in the way many businesses create, manage and distribute content,” said Nick Line, CFC’s chief underwriting officer. His stated aim is clarity rather than reliance on implied or silent cover.

That choice is what makes this more than a product update, because the other major market is moving the opposite way. Verisk subsidiary ISO filed AI exclusion endorsements in the US in January, with AIG and Berkley following across general liability and professional lines. The London market has been slower: the Lloyd’s Market Association’s head of technical underwriting, David Powell, has said the association is waiting for an instruction from the market before drafting equivalent exclusions. Some Lloyd’s syndicates have gone further towards cover — Chaucer and coverholder Armilla offer a combined cyber and standalone AI liability structure with aggregate limits of $25m or more, while Beazley and QBE have introduced AI sublimits capping AI-related payouts at around 10% of policy limits.

Looking Forward

An LMA survey of underwriters put professional indemnity as the line with the highest perceived exposure to AI-related losses, and Lloyd’s now books AI within an innovation risk category accounting for roughly 5% of gross written premium. The claims history behind that is UK-specific: both the Law Society and the Bar Council issued generative AI guidance after a Divisional Court ruling on fabricated citations, with the Bar Standards Board adding more in May, and comparable disputes have appeared in the US, Canada and Australia. For UK brokers advising professional services clients, the affirmative-versus-exclusionary split becomes a renewal comparison rather than a technicality — two policies at similar prices may respond very differently to the same hallucinated citation. Firms carrying silent cover today should establish which side of that divide their programme sits on before finding out at claim stage.