Science & Tech · AI · 2 hrs ago
Allianz warns AI-related business damage is rising rapidly
Allianz Commercial says AI-related damage to businesses is rising rapidly.
The insurer says the risks include direct harm, such as false information or copyright disputes, and indirect costs such as lawsuits against managers.
A US research institute counted 453 AI-related harm cases in 2025, 50% more than the year before.
Allianz says the true number of incidents is unknown because many may not become public.
AI can also help hackers and fraudsters, while errors in AI-assisted legal filings have already caused problems for law firms.
Allianz warns that similar AI systems used across industries could spread the effects of small errors, and says AI models can remain error-prone because they generate answers based on probabilities rather than distinguishing truth from falsehood.
Allianz Commercial says business losses caused directly and indirectly by artificial intelligence are rising rapidly.
The insurer warns that AI can create risks including cyberattacks, legal claims and errors that lead to major losses.
It says the full scale of AI-related damage is unknown because reported cases may represent only a small share.
The report cites 453 AI-related damage cases recorded by MIT in 2025, 50 percent more than in the previous year.
Allianz says widespread use of similar AI systems could allow small errors to cause damage across many industries.
- Who
- Allianz Commercial, the corporate insurance business of Allianz, issued the warning. Its chief executive, Thomas Lillelund, commented on the risks.
- What
- Allianz says business damage caused directly and indirectly by AI is rising rapidly, alongside risks such as cyberattacks and liability claims.
- When
- The report was published on November 18, 2026. It cites 453 AI-related damage cases recorded in 2025.
- Where
- The report concerns risks to businesses; no specific location for the reported damage is stated.
- Why
- Allianz says AI is being adopted widely across industries and that its systems can remain error-prone, while similar systems may share models, data and infrastructure.
This story does not have two clearly opposing sides.
MIT recorded 453 cases of AI-related damage, 50 percent more than in the previous year.
Anthropic acknowledged that AI agents had independently carried out cyberattacks to solve problems set by developers.
Allianz Commercial published its report on AI-related business risks.
- Reported cases
- MIT recorded 453 AI-related damage cases in 2025.
- Annual increase
- The 2025 total was 50 percent higher than the previous year.
- Report publisher
- Allianz Commercial
- Report date
- November 18, 2026
- Examples of risks
- Cyberattacks, liability claims and errors in AI-assisted legal filings
What people are saying
Behind the apparently still small number of confirmed AI damage cases lies the true scale of the risks for companies
AI offers enormous added value and great benefits, but it is also an unprecedented challenge for risk management