Allianz: Confirmed AI Losses Are 'Tip of the Iceberg' as Suits Surge
Allianz Commercial says confirmed AI insurance losses hide the real exposure, with incidents up 150% in two years and single cases potentially costing the industry US$100mn each.
By Amara Osei
4 min read

What's News
- MIT tracked 453 AI incidents in 2025, up from 179 in 2023 — a 150%+ rise in two years.
- 22 AI-related securities class actions were filed in the U.S. by end of August 2026, per Cornerstone Research.
- Allianz says individual AI cases could produce industry losses of around US$100mn each.
- 86% of 250 surveyed D&O professionals in 25 countries expect AI claims to significantly affect the D&O landscape.
- RAND's 2026 report found most insurers remain silent on AI in policy wording.
Allianz Commercial is already tracking AI-related claims worth millions of euros, and it says single cases could each cost the industry around US$100mn — even though confirmed AI losses barely register in claims data today.
The insurer's report, released November 19, 2026, draws on claims data, incident databases and a survey of D&O professionals. Its core argument: AI risk is now a board-level governance challenge, not a technology issue.
How fast are AI incidents and lawsuits rising?
The MIT AI Risk Initiative tracked 179 reported AI incidents in 2023, 307 in 2024 and 453 in 2025, with data as of October 7, 2026. That's a rise of more than 150% in two years, and Allianz sees "no sign of slowing."
Securities class actions tied to AI in the U.S. follow the same curve: seven filings in 2023, 16 in all of 2025, and 22 by the end of August 2026, according to Cornerstone Research and The D&O Diary. The main driver, per Allianz, is "AI washing" — companies overstating AI capabilities in investor disclosures and facing shareholder suits when results disappoint.
IP litigation has turned active too. As of June 2026, artists, entertainment and media companies had filed 115 IP lawsuits against AI developers, per a tracker maintained by ChatGPT Is Eating the World.
Where are the losses hiding?
Based on current claims data, confirmed AI-related losses wouldn't rank among the top 20 loss drivers, Allianz says. The insurer argues that dramatically understates exposure. Most AI claims get absorbed into existing categories — cyber, D&O, product liability, crime — without being flagged as AI losses.
"Today's apparently modest number of confirmed AI claims hides the true extent of exposures for businesses," said Thomas Lillelund, CEO of Allianz Commercial. Michael Daum, head of the AI Risk and Claims Center of Excellence, added: "What we observe today with confirmed AI losses is almost certainly just the tip of the iceberg."
Tracked claims include vehicle accidents tied to alleged defective AI safety systems and shareholder securities litigation over misleading disclosures.
How does Allianz break down AI loss?
The report organizes exposures into four categories with direct implications for policy placement:
- Direct AI loss. The system itself causes harm through a wrong output, decision or recommendation — hallucinated financial advice, a defective vehicle safety system.
- AI-related loss. AI is a material factor, but the loss stems from governance, legal or reputational issues. AI-washing securities suits are the clearest example.
- AI-accelerated loss. AI enables or scales malicious activity. The FBI's 2025 Internet Crime Report attributed roughly $893 million in AI-related cybercrime costs to U.S. citizens; Crowdstrike reported an 89% rise in AI-enabled adversary attacks.
- AI-dependency loss. The initial event isn't AI-related, but reliance on AI-driven processes with no manual fallback amplifies the loss. The report cites a large retailer forced back to paper-based warehouse processes after a third-party technology outage, causing inventory imbalances and significant financial losses.
What makes AI different from earlier tech waves?
AI is being adopted simultaneously across virtually all industries and geographies, and many systems share foundations — the same large language models, training data, cloud infrastructure and data centers.
"One error in a model or training data could result in losses across multiple sectors and thousands of companies all at the same time," said Michael Bruch, global head of Risk Consulting Advisory Services at Allianz Commercial. That concentration creates correlated loss potential with no close equivalent in earlier liability cycles.
Can AI also cut risk?
Yes — when governed properly. Allianz analysis shows early detection can stop a €20,000 cyber loss from becoming a €20 million one, a 1,000x difference. AI-driven fraud detection, predictive maintenance, supply-chain monitoring and compliance surveillance are already delivering measurable risk reduction. Rishi Baviskar, global head of Cyber Risk Consulting at Allianz Commercial, said AI detection models have cut false positives by up to 70% versus legacy systems.
Why the coverage gap matters
A 2026 RAND Corporation report on AI insurability found carriers split three ways: a minority affirmatively cover AI losses, a growing number file broad exclusions, and the majority remain silent. That silence isn't neutral — coverage then depends on how a claim gets characterized after the loss, varying by carrier and line.
In D&O, insurers are watching AI-washing exposure closely but have largely stopped short of broad exclusions, leaving wording effectively silent for most corporate clients. A survey of 250 professionals in 25 countries, conducted by Allianz with The D&O Diary, found 86% believe AI-related claims will significantly affect the D&O landscape, and 92% agree poor AI governance raises claim likelihood.
Allianz's bottom line: leaders must understand where AI is used, how it shapes decisions and which exposures emerge when tools fail — before losses become visible at scale. The next renewal cycle will test whether policy wording catches up.
Source: epyhia.com
More from Amara Osei
Show full bio
Senior reporter covering companies, deals and strategy in entrepreneurship, startups and venture funding at Business Bearings.
2,503 articles