A humanoid robot in Hangzhou, China, now issues traffic warnings without a human operator standing by. Technology firm SUPCON has deployed 15 autonomous robots since May and says they have issued more than 170,000 warnings and cut some violations by more than 40%, though Reuters said it could not independently verify the figures.
It is one pilot among many: China delivered more than 40,000 humanoid robots in the first half of the year alone, according to a Chinese industry body cited by Reuters. For UK brokers, the direction of travel matters more than any single robot. As clients move from supervised systems to genuinely autonomous ones, existing insurance programmes may not automatically flex to cover them.
Andrew Hill, global head of cyber product and innovation for WTW's Financial, Executive and Professional Risks (FINEX) division in London, said the starting point for brokers is the policy wording itself, particularly coverage triggers and exclusions. Autonomous robotics can cut across cyber, professional and technology E&O, property, general liability, products liability and employers' liability.
While many policies will be agnostic as to whether a loss was caused by a person or an autonomous system, some wording still assumes human involvement. With many policies predating the emergence of AI and autonomous robotics, Hill said cover should not be taken for granted.
"The question is not whether a policy responds to 'AI', but whether it responds to the specific loss event that AI could cause," he said.
When a robot causes harm through its own decision-making, rather than a hack or human error, working out which policy responds becomes more complicated. Losses arising from autonomous AI decision-making can still be covered, Hill said, provided they fall within the insuring agreement and are not otherwise excluded.
"The key issues for brokers are whether the policy does specify that a human actor must be responsible for the act or event, which could limit the cover in the context of agentic or autonomous decision making, and whether any AI-specific exclusions apply," he said, adding that such exclusions remain relatively uncommon, but where they exist they may materially alter the coverage position.
A single malfunction can also cross several lines of cover at once, potentially creating coverage gaps or grey areas between policies. Brokers should assess the exposure across the entire insurance programme rather than considering individual policies in isolation, he said.
"Robot malfunctions can give rise to a combination of cyber, bodily injury, property damage, professional liability and product liability exposures. Because many policy wordings were drafted before the widespread adoption of AI and autonomous systems, there is a real possibility that coverage gaps or grey areas emerge between policies," Hill said.
Exclusions deserve particular scrutiny, he added. "For example, could a cyber attack against an autonomous robot that results in property damage be excluded under a property policy because of a cyber exclusion, while also falling outside the scope of a cyber policy because of a property damage exclusion? These are precisely the types of coverage gaps brokers should be identifying."
Cross-border deployment adds a further layer. The regulatory landscape for AI is evolving rapidly and becoming a key area of exposure for organisations deploying autonomous systems, Hill said.
"The EU AI Act, which is now in force, is arguably the most comprehensive AI regulatory framework currently in existence. Importantly, it has extraterritorial reach, meaning organisations developing, supplying or using AI systems in the EU may fall within scope even if they are headquartered elsewhere."
For brokers advising clients on the legal implications of AI for the insurance industry, the task extends beyond traditional liability cover to helping clients understand their compliance obligations and how far their programme responds to regulatory investigations and defence costs, Hill said.
With limited claims history to draw on, establishing the intended coverage position before a loss matters more than usual.
"Avoiding assumptions on coverage is critical," Hill said. "Brokers should work closely with clients to understand how autonomous robots are being deployed, identify the material loss scenarios that could arise, and evaluate those scenarios against the organisation's existing insurance programme. Where uncertainty exists, brokers should engage insurers to clarify intent and, where appropriate, amend policy wording to remove ambiguity before a loss occurs."
Where a genuine gap falls outside insurer appetite, it should be "clearly communicated to the client together with recommendations for mitigation," he said, adding that brokers "may also wish to explore the emerging specialist AI insurance market, which has developed significantly over the last two years."
Dedicated products are already beginning to emerge. In a separate WTW analysis published in August, technology and future liability analyst Sonal Madhok pointed to China Pacific Insurance's "Ji Zhi Bao" policy, launched in 2025, which covers a robot from testing through to deployment, combining property damage and third-party liability cover.
For brokers, however, the arrival of autonomous robots does not necessarily call for an entirely new insurance product. It calls for knowing exactly where a loss would sit across a client's existing programme, and finding the gaps before a real-world loss puts the wording to the test.
