Glossary

Speak the language of AI insurance

Plain-English definitions of the coverage, distribution, and risk terms that come up when AI companies and brokers place generative AI liability cover.

A

6 terms

  • Absolute AI Exclusion

    An insurance endorsement that bars coverage for any claim arising out of the use, deployment, or development of artificial intelligence, with no carve-back.

  • Affirmative AI Coverage

    Policy language that explicitly grants coverage for claims arising from artificial intelligence systems, rather than leaving the exposure ambiguous.

  • Agentic AI Liability

    Third-party liability arising when an autonomous AI agent takes an unauthorized action, transacts on a user's behalf, or causes financial loss outside its intended scope.

  • AI Hallucination Liability

    The third-party legal exposure a deployer faces when a generative AI system produces a false, fabricated, or unsupported output that a user relies on to their financial detriment.

  • AI Indemnification

    A contractual promise from an AI vendor (such as Microsoft Copilot Copyright Commitment or Google Cloud) to defend customers against IP claims arising from the vendor's AI outputs.

  • Algorithmic Bias Liability

    The legal exposure a company faces when an AI system produces discriminatory or disparate outcomes in hiring, lending, housing, or pricing decisions.

C

2 terms

  • Claims-Made Policy

    A policy that responds only to claims first made against the insured and reported to the carrier during the policy period, regardless of when the underlying act occurred.

  • Cyber Liability Insurance (Cyber)

    Coverage for losses arising from cyber-attacks, data breaches, and the regulatory and third-party fallout that follows.

D

1 term

  • Duty to Defend

    An insurer's obligation to fund and provide legal defense for any lawsuit where at least one allegation, if proven, would fall within policy coverage.

F

1 term

  • Foundation Model Liability

    The exposure faced by developers of base AI models (OpenAI, Anthropic, Google) and by enterprises that build downstream applications on those models.

G

2 terms

H

1 term

  • Hammer Clause

    A consent-to-settle provision that caps an insurer's liability at a rejected settlement amount if the insured refuses to settle a claim the insurer recommends resolving.

L

1 term

  • Lloyd's Coverholder

    A company authorized by a Lloyd's managing agent to enter into contracts of insurance on behalf of the members of a Lloyd's syndicate, under a binding authority.

N

1 term

  • NAIC Model AI Bulletin (AIS Program)

    A National Association of Insurance Commissioners bulletin adopted in December 2023 requiring insurers to maintain a written AI governance program (AIS Program) for all consumer-facing AI use.

P

1 term

  • Prompt Injection Liability

    The exposure created when an attacker manipulates an AI system's instructions to exfiltrate data, bypass guardrails, or cause unintended actions on behalf of a user or company.

R

1 term

  • Retroactive Date

    The earliest date on which a wrongful act can have occurred and still be covered under a claims-made policy, anchoring how far back prior acts are insured.

S

2 terms

  • Silent AI

    A coverage gap where an existing insurance policy is neither affirmatively granting nor explicitly excluding AI-related claims, leaving the insurer's exposure undefined.

  • Surplus Lines Insurance (E&S)

    Specialty insurance written by non-admitted carriers for risks the admitted market will not insure, regulated state-by-state with surplus lines taxes.

T

3 terms

W

1 term

  • Wrongful Act

    The defined event that triggers coverage under a claims-made liability policy, typically an actual or alleged act, error, or omission in the insured's covered activities.