People are skeptical of agents in general, and the major risk is that companies don't test agents enough.
Driving the news: AXA XL, the specialty risk division for property and casualty insurance of AXA, has introduced a new extension to its cyber insurance policies that specifically addresses risks related to developing generative AI models, the company said in a press release. The move marks a significant step as insurers adapt to emerging risks from AI technologies that have become critical in business operations.
Zooming in: The new coverage functions as an add-on to existing cyber insurance policies, providing coverage tailored to the risks involved in building and deploying Gen AI models.
- Data Poisoning: Covers costs from AI model sabotage, including data restoration and litigation expenses.
- Usage Rights Infringement: Protects against claims over unlicensed use of copyrighted materials or IP in AI training.
- Regulatory Violations: Addresses fines or legal costs related to non-compliance with regulations like the EU’s AI Act.
Autonomous risks: Agentic AI poses an additional set of risks, allowing AI to work almost entirely autonomously with a system of checks and balances imposed by other AI "agents". Brooke Hopkins, a former tech lead at Waymo and current Founder of YC'24 company Coval, is building evaluation systems for voice and chat agents. "The way we define it, there are really two types of agents: there are internal agents working within their own systems, and then there are autonomous agents interacting with external systems, external users, etc," says Hopkins. "People are skeptical of agents in general, and the major risk is that companies don't test agents enough."
Why it matters: AXA XL’s new coverage highlights the sheer size of a market that barely existed a few years prior. With the Gen AI market projected to grow from $40 billion in 2022 to $1.3 trillion over the next decade, insurers are scrambling to adapt their products to keep pace, with firms like Aon and Marsh also launching AI-focused services. The quick pivot underscores AI’s dual role as both an industry disruptor and a risk driver.
