Chaucer is a specialty insurer at Lloyd's and part of China Re Group. It co-developed Armilla's AI liability policy and launched Vanguard AI with Armilla in February 2026.
Two-part structure from Chaucer and Armilla that pairs Chaucer's cyber and tech E&O with Armilla's standalone AI liability policy under agreed rules for splitting mixed losses. The aim is to stop AI-driven claims from eating into cyber or tech E&O limits.
Item 1
Published limit
$25M+
AI aggregate, alongside $10M cyber
Item 2
Type
Program
Item 3
For
Companies that use AI
Item 4
How to buy
Surplus lines brokers
Item 5
Backed by
Chaucer (China Re Group member, via Lloyd's and company markets) for cyber/tech E&O; Armilla AI policy underwritten as Lloyd's coverholder
Item 6
Who it is for
Organisations deploying AI at scale, including generative AI and customer-facing agents
Item 7
Where
Not stated
Item 8
Launched
10 February 2026
Item 9
Risks named
Hallucinations and model errors, AI agent errors, AI-driven cyber events
Item 10
Record
Confirmed by issuer
What the sources don't say: Chaucer's release gives a $25M+ AI aggregate alongside $10M of cyber cover; the same sentence in Armilla's release is garbled, so the limit shown follows Chaucer's version. The sources do not state a territory for this product.
Cowbell's non-admitted cyber product for larger US businesses explicitly picks up AI-related incidents alongside quantum-computing risks and a full cyber coverage base. The reported maximum limit is $10 million.
Item 1
Published limit
$10M
maximum limit
Item 2
Type
Standalone policy
Item 3
For
Companies that use AI
Item 4
How to buy
Not stated
Item 5
Backed by
Chaucer Insurance Company, non-admitted
Item 6
Who it is for
Organisations with annual revenue of $250 million to $1 billion
Item 7
Where
United States (non-admitted)
Item 8
Launched
21 April 2026
Item 9
Risks named
AI-driven cyber events
Item 10
Record
Confirmed by issuer
What the sources don't say: Cowbell's release says the AI cover includes unauthorised use of or access to AI systems, but it gives no state-by-state availability.
Hallucinations and model errors: Vanguard AI (002)
AI agent errors: Vanguard AI (002)
AI-driven cyber events: Vanguard AI (002) and Prime One (013)
Risk names are our labels for what each source describes. Wording, sublimits and exclusions decide what a policy pays.
Change history
Cowbell Prime One adds affirmative AI cover
A non-admitted cyber product written by Chaucer for US firms with $250M to $1B revenue that affirmatively covers AI-related incidents, with limits up to $10M.
Yes. The register lists 2 offerings from Chaucer: Vanguard AI (program) and Prime One (standalone policy). Each is confirmed on the issuer's own pages.
What limits does Chaucer publish?
Vanguard AI: $25M+, AI aggregate, alongside $10M cyber; Prime One: $10M, maximum limit.
How do you buy Chaucer's AI cover?
Vanguard AI: Surplus lines brokers. We do not sell insurance or take commissions.
