Lloyd's of London excluded AI liability from standard policies in 2023. There is currently no actuarial basis for pricing AI risk. Companies deploying AI at scale are self-insuring against unknown exposures.
$0Commercial AI liability insurance market (2023)
$500B+Estimated uninsured AI liability exposure globally
2023Year Lloyd's excluded AI from standard policies
3–5yrEstimated timeline for insurance market to develop
Choose your depth. The data doesn't change — just the explanation.
Insurance companies make money by calculating how likely something bad will happen and charging you money to protect against it. But with AI, nobody knows how often AI will cause accidents or how much damage it will do. So insurance companies said "we won't cover AI problems" — leaving businesses that use AI to pay for any mistakes themselves.
Insurance requires actuarial data — historical loss rates to price risk. AI liability is new, varied, and often ambiguous (who is liable when AI causes harm — the developer, the deployer, or the user?). Lloyd's of London Bulletin Y5381 (2023) clarified that AI-generated cyberattacks and AI errors are excluded from standard cyber and professional indemnity policies. Some specialty markets offer limited AI coverage, but at very high premiums and low limits. Most enterprise AI deployers currently have coverage gaps of hundreds of millions of dollars in theoretical liability.
Lloyd's Market Bulletin Y5381 (August 2023): AI exclusions in cyber, E&O (Errors & Omissions), and professional indemnity policies. The actuarial gap: no historical frequency/severity data for AI liability claims (too new). Alternative risk transfer emerging: captive insurance (companies self-insuring through subsidiary), parametric insurance (payout triggered by defined AI incident metrics), AI-specific warranties in enterprise contracts. EU AI Act liability provisions: mandatory liability insurance for high-risk AI systems — but no market to buy it yet. UK Law Commission: "AI and the Law" (2024) recommends strict liability framework for high-risk AI.
Lloyd's Bulletin: lloyds.com/news-and-insights/market-bulletins/bulletins/2023/08/y5381. EU AI Act liability: Article 78+, Annex III (high-risk AI systems). UK Law Commission report: lawcom.gov.uk/project/automated-vehicles-and-liability (adjacent). Alternative risk: Marsh McLennan "AI Risk Report" 2024, Munich Re AI liability whitepaper (2023). Track: ISO/IEC 42001 (AI management standard) adoption as proxy for insurable AI deployments; NAIC (National Association of Insurance Commissioners) AI working group (naic.org/ai). Emerging market: Coalition, Beazley, AXA XL specialty cyber with AI riders.
The Insurance Gap
Traditional insurance markets price risk using historical data. AI liability has no historical data. The result: companies deploying AI are exposed to unquantified, uninsured liability.
AI Liability Exposure vs. Available Insurance Capacity ($B)
Estimated enterprise AI liability exposure vs. specialty market capacity. 2024.
Medical AI Error
$1–50M
AI misdiagnosis or wrong treatment recommendation. Per incident. Clinical context only.
Autonomous Vehicle
$0.5–10M
AI-caused accident with personal injury. Traditional auto liability applies inconsistently.
Financial AI Error
$10–500M
Algorithmic trading failure, AI-generated bad financial advice. Varies by scale.
AI hiring/lending bias. Multiple plaintiffs. EEOC enforcement. Unlimited potential.
Critical Infrastructure
$100M–1B+
AI failure in power grid, water system, hospital. Potentially catastrophic.
Insurance Market Development Timeline
Estimated timeline for AI liability insurance market maturation.
AI Risk Categories Excluded by Standard Policies
Coverage gaps in standard commercial policies (2024).
The Self-Insurance Gap
With no commercial insurance available, companies deploying AI at scale are effectively self-insuring — betting that AI errors won't trigger catastrophic liability. For small and mid-sized companies with weak balance sheets, a single major AI incident could be existential. This is the risk that Lloyd's recognized in 2023 that most AI companies haven't.
Sources
• Lloyd's of London. Market Bulletin Y5381. lloyds.com. August 2023.
• Marsh McLennan. "AI Risk: A New Insurance Challenge." 2024.
• Munich Re. "Liability for AI." munichre.com. 2023.
• EU AI Act. Articles 78+ on AI liability. eur-lex.europa.eu. 2024.
• NAIC. AI Working Group Reports. naic.org/ai. 2024.