What Did the Insurance Industry Reveal About AI Risk?

Major insurers are refusing to cover AI-related liabilities because they see systemic risk they cannot model. At the same time, the AI industry is lobbying for federal laws that override state consumer protections. The result is a liability shield where companies deploy AI at scale, but you bear the risk when failures occur.

Video – “Terrified Of AI” – Insurance Giants PANIC As AI ERRORS Cost Companies Billions

Core Answer:

• Insurers like AIG and WR Berkley are excluding AI from corporate policies because AI introduces correlation risk at unprecedented scale
• One AI failure affects thousands of companies simultaneously, breaking traditional insurance models
• Federal preemption eliminates state-level AI regulations before they establish consumer protections
• 95% of companies use AI, yet almost none have insurance coverage for it
• The liability gap incentivizes rapid deployment over adequate testing

When the people who insure oil rigs refuse to insure your technology, pay attention.

Major insurers including AIG, Great American, and WR Berkley are petitioning U.S. regulators to exclude AI-related liabilities from corporate policies. Not limit coverage. Exclude it entirely.

These are companies that underwrite nuclear plants, space launches, and offshore drilling platforms.

They have actuarial tables for asteroid strikes and pandemic outbreaks. They know how to price catastrophic risk.

And they will not touch AI.

Why Traditional Insurance Models Break Down With AI

An Aon executive identified the structural problem: “We handle a $400 million loss to one company. What we cannot handle is an agentic AI mishap that triggers 10,000 losses at once.”

Traditional insurance relies on historical data of human error. Humans make mistakes at predictable rates. Humans operate at human speed. Humans fail independently.

AI systems fail differently.

AI makes errors at scale and speed humans cannot match. When one model fails, the cascade spreads across industries in ways traditional risk models never anticipated.

Many companies rely on the same foundational AI models from OpenAI, Google, or Anthropic.

One upstream failure. Thousands of downstream losses. Simultaneously.

The insurance industry sees what most executives miss. AI introduces unprecedented correlation risk. The efficiency that makes AI attractive creates systemic vulnerability.

What This Means: AI creates correlated risk that breaks the fundamental assumptions of insurance pricing.

Video – Insureres Wont Cover AI?

How Companies Are Testing Liability Boundaries

Air Canada argued its chatbot was “a separate legal entity responsible for its own actions” after the bot gave incorrect bereavement fare information. The British Columbia Civil Resolution Tribunal rejected this defense.

The attempt tells you where this is headed. Companies are testing whether AI creates a liability shield.

If your AI acts outside your control, are you responsible? If your AI was trained on data you did not create, who owns the liability for outputs? If your AI interacts with another AI and produces harm, where does accountability land?

These questions are being litigated right now.

Insurers are watching companies try to offload responsibility onto algorithms. And they are walking away from the entire category.

What This Means: The legal framework for AI liability is undefined, creating uncertainty that insurers refuse to underwrite.

The Federal Strategy to Eliminate State Protections

While insurers retreat, the AI industry executes a different strategy.

President Trump signed an executive order on December 11, 2025, establishing an AI Litigation Task Force within the Department of Justice to challenge state AI laws.

The order directs the Attorney General to sue states over AI regulations and blocks states with targeted regulations from receiving BEAD funding.

This is regulatory capture operating in real time.

Large AI companies and their trade associations requested federal preemption of state AI laws in submissions to the White House AI Action Plan.

Following these requests, Big Tech companies and their trade groups lobbied for the moratorium in the reconciliation bill.

The timing matters. The insurance crisis reveals that we are deploying AI without understanding its harms.

The AI industry response is not to slow deployment. The response is to eliminate regulatory frameworks that might constrain it.

Federal preemption gets framed as creating “clarity.” What preemption does is eliminate stronger state-level consumer protections before they establish precedents.

What This Means: AI companies are using federal policy to prevent states from creating liability frameworks that protect consumers.

What Uninsurable Risk Does to Deployment Decisions

Verisk, one of the largest creators of standardized policy forms in the U.S. insurance market, introduced new general liability endorsements effective January 2026.

That allow carriers to exclude generative AI exposures. 95% of U.S. companies use generative AI according to a Bain & Company report.

Read that data point again. Nearly every company uses AI. Almost no company has insurance for it.

This creates implicit liability protection for AI deployment. If you cannot buy insurance, you cannot transfer risk.

If you cannot transfer risk, you either stop deploying or accept that losses land directly on your balance sheet.

Most companies choose deployment.

This is how inadequately tested AI systems reach production. Not because companies are reckless. Because economic incentives favor speed over safety.

If insurance will not cover you either way, there is no financial reason to slow down.

The pharmaceutical industry offers a precedent. When liability protections concentrate, costs shift to consumers.

When federal rules override state protections, incumbents entrench advantages. When industries shape legislation proactively, public interest becomes an afterthought.

What This Means: The absence of insurance removes the financial incentive for thorough AI testing before deployment.

Who Bears the Risk When Systems Fail

Insurance retreat combined with federal preemption combined with deployment acceleration equals systemic risk without accountability mechanisms.

The insurance industry says they cannot model the risk. The AI industry says they do not want external regulation. The federal government says state protections are obstacles.

This is not about fostering technology development. This is about who bears the cost when things break.

Right now, that answer is you.

Companies deploying AI are protected by the absence of insurance requirements.

Companies building AI are protected by federal preemption of state laws. Insurance companies are protected by exclusions they write into every policy.

The only unprotected party is the one who will experience the consequences of AI failures at scale.

When professional risk assessors refuse to price a technology, and the industry building that technology lobbies to eliminate oversight, you are not watching progress. You are watching the construction of a liability shield.

The question is not whether AI will fail at scale. The question is who pays when failure occurs.

Right now, the answer gets written into federal policy and insurance exclusions. And you are not part of that conversation.

Frequently Asked Questions

Why are insurance companies refusing to cover AI risks?
Insurance companies rely on historical data to model risk. AI introduces correlation risk where one failure affects thousands of companies simultaneously, breaking traditional actuarial models. Insurers cannot price something they cannot model.

What is federal preemption of AI laws?
Federal preemption means federal AI regulations override state laws. AI companies are lobbying for this because it prevents individual states from creating stronger consumer protections or liability frameworks that might constrain AI deployment.

What percentage of companies use AI without insurance coverage?
According to Bain & Company, 95% of U.S. companies use generative AI. Verisk introduced endorsements in January 2026 allowing insurers to exclude AI exposures, meaning most companies operate without coverage for AI-related liabilities.

How does Air Canada’s chatbot case relate to AI liability?
Air Canada tried to argue its chatbot was a separate legal entity responsible for its own actions after it provided incorrect fare information. The court rejected this defense, but the attempt shows companies are testing whether AI creates liability shields.

What happens when companies deploy AI without insurance?
When insurance is unavailable, companies cannot transfer risk. This removes the financial incentive for thorough testing because there is no insurance premium discount for safer practices. The result is faster deployment of less-tested systems.

Why does the insurance industry’s refusal matter to consumers?
When insurers refuse to cover AI, it signals they see risks they cannot quantify. Combined with federal efforts to eliminate state regulations, this creates a gap where companies deploy AI at scale, but consumers bear the consequences when failures occur.

What is correlation risk in AI systems?
Correlation risk means multiple failures happen simultaneously rather than independently. Because many companies use the same foundational AI models, one upstream failure cascades to thousands of downstream companies at once.

How does this compare to other emerging technologies?
Insurance companies routinely cover high-risk technologies like nuclear power and space launches because they have data to model the risks. AI is unique because the risk profile is unknown and correlated across the entire economy.

Key Takeaways

• Professional risk assessors who insure nuclear plants and space launches refuse to underwrite AI because correlation risk breaks traditional insurance models
• AI companies are lobbying for federal preemption of state laws to eliminate consumer protections before legal precedents form
• 95% of companies deploy AI while almost no company has insurance coverage for AI-related liabilities
• The absence of insurance removes financial incentives for thorough testing, accelerating deployment of inadequately tested systems
• When AI fails at scale, liability falls on consumers rather than the companies deploying or building the technology
• The combination of insurance retreat, federal preemption, and deployment acceleration constructs a liability shield for AI companies
• The insurance industry’s refusal to price AI risk is a market signal that the technology’s harm profile is not understood