Artificial intelligence: loss scenarios and claims developments
This report looks at how AI is changing the risk and claims landscape, why losses may lag behind adoption, and highlights what businesses can do to better understand, manage and transfer AI-related risks.
Artificial intelligence (AI) is transforming the global economy at unprecedented speed, moving rapidly from emerging technology to a boardroom priority. Adoption is accelerating and AI is already embedded in many business processes – from production planning to autonomous systems to pricing models.
In just one year, AI has also surged from #10 to #2 in the Allianz Risk Barometer – the largest year-on-year rise of any category in the annual business survey. This dramatic shift reflects growing awareness among businesses and insurers that AI is not simply a technology issue but a fundamental business, governance, and leadership challenge.
Yet despite this heightened concern, confirmed AI‑related insurance losses remain relatively low. Based on current claims data, AI would not rank even in the top 20 loss drivers. However, AI losses are frequently underreported, difficult to quantify or attribute, or are recorded under traditional categories such as cyber, directors’ and officers’ (D&O) liability, or product liability insurance. The true risk exposure is substantial and growing fast. A sharp increase in losses may be anticipated.
AI incidents are accelerating
AI-related incidents and litigation have risen in recent years with corporates being sued for a wide range of AI-related issues, including intellectual property and copyright infringement, data privacy breaches, hallucinations (false information), discrimination, bias and environmental issues, as well as the overstatement or misrepresentation of companies’ strategic capabilities.
Databases tracking AI incidents and lawsuits show they are significantly increasing every year. For example, the number of incidents rose from 179 in 2023 to 307 in 2024 and 453 in 2025, according to the MIT AI Risk Initiative (as of October 7, 2026) and this trajectory shows no sign of slowing.
At the same time, AI-related securities class action lawsuits in the US have followed a similar pattern, with 22 filings in 2026 alone (as of the end of August), compared to 16 in all of 2025.
Allianz Commercial is already tracking AI-related insurance claims with substantial values in the millions of euros. Examples of cases include accidents caused by alleged defective AI safety systems in vehicles and shareholder-filed securities litigation alleging misleading AI-related disclosures and omissions by company executives. Such cases could easily result in total insurance industry losses in the region of US$100mn each.
Four categories of AI loss
AI risk operates on several levels. One level lies within the technology itself. AI systems rarely express uncertainty. At the same time, they are highly sensitive to prompts and can exhibit confirmation bias, meaning small nudges can significantly influence outcomes. This can cause issues in negotiations, assessments or decision-support scenarios.
A second level arises from how AI is used in organizations. Experience is still limited across many functions. At the same time, competitive pressure to deploy AI quickly is enormous. Even when companies act responsibly, the probabilistic nature of AI means that mistakes could still happen. A further level consists of indirect risks. In many cases, AI is not the direct cause of damage, but a force multiplier. For risk management purposes, it is therefore crucial to interpret AI-related harm broadly.
The report introduces a practical framework for understanding how AI can generate losses and claims, organized into four main categories:
Direct AI loss
The AI system itself is the primary cause of harm through an incorrect action, decision, recommendation or output.
AI-related loss
AI is a material factor in the loss event, but the loss arises from surrounding business, governance, legal, reputational, or third-party AI issues rather than the insured’s own AI system.
AI-accelerated loss
AI does not cause the loss directly but enables, scales, or accelerates malicious activity.
AI-dependency loss
AI does not cause the initial event, but heavy reliance on AI-driven processes and insufficient fallback capabilities significantly increase the magnitude, duration, or cost of the loss.
A unique risk profile
AI carries a unique systemic risk profile. Unlike previous technology revolutions, AI is being adopted simultaneously across virtually all industries and geographies. Many AI systems share common foundations – the same large language models, training data, cloud infrastructure, and data centers. This concentration means that a single flaw, outage, or vulnerability could cascade across sectors and borders. AI also operates probabilistically, meaning errors are inevitable even in well-designed systems. When deployed at scale, even a small error rate can generate a massive number of incidents simultaneously.
AI as a risk management tool
While AI introduces new exposures, it can simultaneously reduce the frequency and severity of existing business risks when deployed with appropriate governance, controls, and human oversight. Many practical examples already demonstrate AI’s risk reduction potential. In financial services, AI-driven fraud detection systems identify suspicious transactions in real time, helping prevent payment fraud, account takeovers, and money laundering. In cyber security, AI is increasingly used to detect abnormal network activity, identify compromised credentials, and uncover emerging attack patterns before significant damage occurs.
Manufacturing companies use predictive maintenance models to detect early signs of equipment failure. Supply chain organizations employ AI to monitor suppliers and logistics networks, enabling earlier identification of disruption risks and more resilient business continuity planning. AI is also enhancing regulatory compliance by continuously monitoring large volumes of transactions, communications, and records, helping organizations identify potential compliance breaches at an earlier stage and reducing the likelihood of regulatory action.
Organizations using AI-driven prevention and detection tools have materially lower breach and business interruption costs. The cost of a cyber breach that progresses to data theft and encryption can be 1,000 times higher than an incident that is detected and contained early.
The bottom line
For business leaders, the most important shift is to treat AI risk not only as an IT issue, but as a business, governance, and leadership challenge. Organizations need to understand where AI is being used, how it affects decision-making and operations, and which exposures could emerge if tools fail, produce inaccurate outputs, or amplify existing risks. This requires deliberate decisions about which risks can be mitigated, which can be accepted, and which should be transferred. It also requires stronger visibility across functions, because AI risks can sit at the intersection of technology, legal, compliance, operations, product liability, cyber security, and board accountability.
The opportunities for AI remain significant. But realizing them sustainably depends on understanding the risk landscape early and proactively, before losses become more visible at scale.
Photo: Shutterstock
Artificial intelligence: loss scenarios and claims developments webinar | December 7, 2026
Join our leading experts as they discuss emerging AI loss scenarios and claims trends, including data privacy breaches, copyright infringement, AI washing, and other evolving liability exposures.
Drawing on real-world examples and claims insights, the session will explore how AI risks are materializing across industries, why many AI-related losses remain difficult to identify, and what organizations can do to better understand and manage these exposures. A live Q&A session will follow the presentation.
If you are interested to join the webinar, register via the link below:
Our experts
Allianz Commercial's insurance experts around the world provided their insights and knowledge to this report.