Major Insurers Retreat from AI Coverage as Multi-Billion Dollar Risk Concerns Mount

Insurance Giants Pull Back from AI Coverage

Major insurance companies like AIG, WR Berkley, and Great American are seeking regulatory permission in the U.S. to exclude AI-related liabilities from corporate policies. This move stems from the escalating and unpredictable financial risks posed by AI technologies, which traditional insurance models struggle to quantify 1

1

2

2

. WR Berkley’s proposed exclusion, for instance, broadly covers any product or service "incorporating" AI, indicating deep concern over widespread liability.

High-Profile AI Failures Spur Caution

This industry shift follows several high-cost AI blunders. Google is facing a $110 million defamation lawsuit after its AI Overview falsely accused a company of being sued. Air Canada was compelled by a tribunal to honor a discount invented by its chatbot. Additionally, a UK engineering firm, Arup, lost HK$200 million (approximately $25 million) to fraudsters who used AI-cloned voices and video of a senior manager to authorize transfers 1

1

2

2

. These incidents highlight the tangible and substantial financial risks associated with current AI applications.

The "Black Box" of AI and Systemic Risk

Insurance experts find AI particularly challenging to underwrite due to its opaque nature. Dennis Bertram from Mosaic Insurance described it as "too much of a black box," making it difficult to assess risk or assign liability 2

2

3

3

. Rajiv Dattani of the Artificial Intelligence Underwriting Company noted, "Nobody knows who's liable if things go wrong" 2

2

4

4

.

Beyond individual claims, insurers are deeply worried about systemic, correlated losses. Kevin Kalinich of Aon warned that while a single company's AI failure causing a $400-500 million loss is manageable, the industry cannot absorb scenarios where a single upstream AI mistake triggers thousands of simultaneous losses across many clients 1

1

2

2

. Ericson Chan of Zurich Insurance highlighted that unlike traditional tech errors where responsibility is clear, AI risk involves multiple parties and could have an "exponential" market impact 2

2

.

Navigating Limited AI Coverage

Despite the widespread exclusions, some insurers are offering niche, limited AI coverage. QBE, for example, has introduced coverage for fines under the EU AI Act, though with significant caps. Chubb provides coverage for certain AI incidents but explicitly excludes widespread, simultaneous events. Insurance brokers caution businesses to carefully review these endorsements, as they may offer less protection than they seem, potentially shifting more risk onto companies' own balance sheets 1

1

2

2

.