- Dotika
- Posts
- Insurers shy from AI settlement costs
Insurers shy from AI settlement costs
ALSO : AI Opens 3D Modeling to Blind Coders
Alexandre HOTTON
October 08, 2025
News of the day
1. AI firms like OpenAI and Anthropic are struggling to get insurers to cover massive potential lawsuit costs → Read more
2. A11yShape merges OpenSCAD and GPT-4o, letting blind programmers create 3D models through code with AI-generated audio descriptions.→ Read more
3. Anthropic plans to open an office in India next year, citing high demand for AI tools and India's significant technical talent → Read more
4. Samsung's new Tiny Recursive Model (TRM) with only 7M parameters achieves state-of-the-art results on complex reasoning benchmarks → Read more
Our take
Hi Dotikers!
The artificial intelligence industry just discovered an unexpected problem: nobody wants to insure it.
OpenAI and Anthropic, two of the sector's most prominent companies, are facing lawsuits that could cost them billions of dollars. And here's the catch: their insurance will only cover a tiny fraction of these astronomical sums.
OpenAI has insurance coverage of around $300 million for AI-related risks, but that's a drop in the bucket compared to potential claims worth several billions. It's like having $10,000 in home insurance to protect a castle.
Why is this happening? Insurance companies are terrified by the scale and unpredictability of AI risks. Anthropic just agreed to pay $1.5 billion to settle a class action lawsuit with authors who accused them of using pirated books to train their models. OpenAI, meanwhile, is being sued by the New York Times for copyright infringement, and even faces a wrongful death case involving ChatGPT.
So what's the workaround? Both companies are now considering using money directly from their investors to pay settlements, a form of "self-insurance." It's a bit like having to pay for your car repairs out of pocket because no insurer wants to cover you.
This crisis reveals that AI is moving faster than the legal and financial guardrails meant to regulate it. Insurers, who are usually willing to cover anything for the right price, are backing away from risks they deem impossible to calculate. This could speed up the arrival of new regulations to clarify legal liabilities related to AI.
A.
Tweet of the day
Our new Gemini 2.5 Computer Use model is now available in the Gemini API, setting a new standard on multiple benchmarks with lower latency. These are early days, but the model’s ability to interact with the web – like scrolling, filling forms + navigating dropdowns – is an
— Sundar Pichai (@sundarpichai)
9:03 PM • Oct 7, 2025
Reply