A.I. Losses Are Just ‘Tip of the Iceberg,’ Allianz Commercial Warns
Confirmed A.I.-related insurance losses would not rank among the top 20 loss drivers based on current claims data, even as A.I.-related incidents tracked globally rose to 453 in 2025, from 307 in 2024 and 179, according to Allianz Commercial. However, the insurer said, many A.I. losses are underreported, hard to quantify, or buried within traditional categories such as cyber, directors’ and officers’ (D&O) liability, or product liability insurance.
Allianz Commercial said it is already tracking A.I.-related claims worth millions of dollars, including vehicle accidents tied to alleged defective A.I. safety systems and shareholder securities litigation over misleading A.I. disclosures, cases the insurer said could each result in total industry losses around $100 million.
A.I. jumped from 10th to second place in the Allianz Risk Barometer this year, the largest year-on-year rise of any category, the company said.
A Four-Part Framework for A.I. Losses
Allianz Commercial organized emerging A.I. exposures into four categories:
- Direct A.I. loss, where a system’s own faulty action or output causes harm.
- A.I.-related loss, where A.I. is a factor but the damage stems from surrounding governance, legal or reputational issues.
- A.I.-accelerated loss, where A.I. scales malicious activity such as fraud or cyber-attacks.
- A.I.-dependency loss, where reliance on A.I.-driven processes worsens the severity of an unrelated event.
Intellectual property disputes have dominated direct A.I. litigation, with 115 lawsuits filed against A.I. developers by artists, entertainment and media companies as of mid-2026, the report said. Law firms have also faced court sanctions for A.I. “hallucinations,” including fabricated citations, while a UK survey found half of accountants reported clients suffering direct financial losses from incorrect tax or financial advice generated by general-purpose A.I. chatbots.
D&O insurance has emerged as one of the most visible fronts, driven by “A.I. washing,” or the overstatement of A.I. capabilities. A.I.-related securities class action filings in the U.S. rose to 16 in 2025, up from seven in 2023, and reached 22 by the end of August 2026, putting the year on pace to nearly double 2025’s total.
A survey of 250 professionals in 25 countries, conducted by Allianz Commercial with The D&O Diary, found 86% believe A.I.-related claims will significantly affect the D&O landscape and 92% agree poor A.I. governance raises the likelihood of claims, with A.I. washing and data privacy each cited by 28% of respondents as the top exposure.
On the accelerated-loss side, attacks by A.I.-enabled adversaries rose 89%, according to cybersecurity firm Crowdstrike, while the FBI said reported A.I.-related cyber crime cost U.S. citizens $893 million in 2025.
Dependency losses have also surfaced: a large retailer hit by a cyber outage at a third-party technology provider was forced to revert to paper-based warehouse processes after losing access to A.I.-enabled forecasting and inventory tools, causing incorrect product selection and operational delays, the report said.
Why the True Scale Remains Hidden
Michael Daum, head of the AI Risk and Claims Center of Excellence at Allianz Commercial, said confirmed A.I. losses today are “almost certainly just the tip of the iceberg,” since many incidents go undetected or unreported and most insurers are not yet equipped to tag claims as A.I.-related.
The report said a large gray area exists where A.I.’s role in a loss cannot be conclusively verified, and the boundary between an A.I. claim and a traditional one, such as privacy or D&O, is often unclear.
Michael Bruch, global head of Risk Consulting Advisory Services at Allianz Commercial, said A.I.’s systemic risk differs from past technology shifts because many systems share common large language models, training data and infrastructure, meaning “one error in a model or training data could result in losses across multiple sectors and thousands of companies all at the same time.”
The report also noted A.I.’s mitigation potential, citing Allianz Commercial analysis showing early detection can stop a 20,000-euro cyber loss from becoming a 20-million-euro one, and that A.I.-driven detection models have reduced false positives by up to 70% compared with legacy systems, according to Rishi Baviskar, global head of Cyber Risk Consulting at Allianz Commercial.
Obtain the full report here. &

