The short version
- AI does not have its own line of insurance yet for most businesses. A loss caused by AI lands on whichever policy matches what happened: a breach, a professional mistake, an injury, or an employment decision.
- Traditional policies were not built for AI. Munich Re has warned that AI risk sitting silently inside traditional policies can leave businesses "without financial protection," and that the risk of AI underperforming was not considered when those policies were priced.
- Exclusions are arriving. ISO, which writes many standard policy forms, has filed optional generative AI exclusions for commercial general liability, and trade press has reported broad AI exclusions on some D&O and E&O policies. See AI exclusions in business policies.
- Affirmative AI coverage now exists. Coalition added an affirmative AI endorsement to its cyber policies (2024), Armilla launched a Lloyd's-backed AI liability policy that covers "hallucinations or inaccuracies" (2025), and HSB, a Munich Re company, introduced AI liability coverage for small businesses (2026).
- Courts hold businesses responsible for their AI. A tribunal held Air Canada liable for its chatbot's wrong advice, and a federal court has let an AI hiring discrimination case proceed against the software vendor itself.
Which policy responds when AI causes a loss
| What happened | Policy most likely to respond | Common gap |
|---|---|---|
| AI tool gives a client or customer wrong information that costs them money | Professional liability (E&O), technology E&O | New AI exclusions; policies that cover only "professional services" as defined |
| AI system is hacked, or causes a data breach | Cyber | Coverage depends on how the policy defines a security failure |
| Deepfake voice or video tricks staff into wiring money | Cyber (social engineering or funds transfer fraud), crime | Low sublimits; conditions on how the fraudulent instruction arrived |
| AI hiring or screening tool is accused of discrimination | EPLI | Claims against the vendor, and the vendor's indemnity, are separate questions |
| AI-generated content infringes copyright or defames someone | General liability (personal and advertising injury, for copyright only in your advertising), media liability | ISO's optional generative AI exclusions can remove this |
| AI-enabled product injures someone or damages property | General liability, products liability | ISO's optional generative AI exclusions can remove this |
| Shareholders sue leadership over AI statements or strategy | D&O | Reported absolute AI exclusions on some D&O forms |
Details by situation: does cyber insurance cover AI, when your AI gives a wrong answer, and AI hiring tools and discrimination.
Coverage written specifically for AI
| Insurer | What it is | What it covers |
|---|---|---|
| Coalition | Affirmative AI endorsement on its U.S. surplus and Canada cyber policies (March 2024) | AI-caused security failures, and funds transfer fraud using deepfakes or other AI |
| Armilla, with Lloyd's underwriters including Chaucer | AI liability policy (April 2025) | AI that fails to perform as intended, including "hallucinations or inaccuracies leading to damages" |
| Munich Re aiSure | AI performance insurance (first policy 2018) | AI models that underperform, including vendor performance guarantees and discrimination claims |
| HSB, a Munich Re company | AI liability for small and mid-sized businesses (March 2026) | Lawsuits tied to a business's AI use, added to partner carriers' business policies pending regulatory approval |
AI coverage is also showing up inside technology E&O. Some tech E&O policies now offer optional AI cover with separate choices for algorithmic bias, intellectual property, training data, and regulatory defense, usually as a sublimit; the ones we see run from $500,000 to $2 million. For a software or AI company, that can be the simplest way to cover a wrong output. See tech E&O insurance.
These products are new and availability varies by state and carrier. We check what your current policies say about AI first, then whether a specific AI product is worth adding.
What to have ready for an AI coverage quote
Insurers pricing AI cover ask about how the AI is built and controlled, not only about revenue. Having these answers ready speeds up the quote:
- What the AI does: the product or service, who relies on its output, and what happens if it is wrong
- Which models you use: third-party providers (such as OpenAI or Anthropic), open-source models, or your own
- Training data: whether you train or fine-tune models, and where the data comes from
- Autonomy: whether a person reviews outputs before they reach customers, or the system acts on its own
- Governance: testing, monitoring, bias checks, and who is responsible for AI decisions
- Contracts: what your client agreements promise about AI, and any AI clauses clients have asked you to sign
- History: any complaints, claims, or incidents involving AI outputs
Who should look at AI coverage now
- Technology companies and IT providers that build or resell AI features
- Consultants, agencies, accountants, and law firms that use AI in client work
- Businesses with customer-facing chatbots or AI-written customer communications
- Employers using AI to screen, rank, or interview candidates
- Health, wellness, and fitness businesses running consumer health data through AI tools, which Washington's My Health My Data Act treats as processing (HIPAA-covered patient records are exempt)
What to do at your next renewal
- Search each policy and endorsement for "artificial intelligence," "generative," and "machine learning."
- Ask your broker whether any AI exclusion was added at renewal.
- Confirm your E&O covers the services you deliver with AI, not only those you deliver by hand.
- Check whether contracts you sign ask about AI use. Our contract insurance checker flags contracts that mention AI.
- Review the laws that may apply: AI laws for Washington and Idaho businesses.