Companies rushing to deploy artificial intelligence may be adding a powerful tool to their networks that cybercriminals can exploit after gaining access, creating a new area of scrutiny for insurance brokers.
Lauren Winchester (pictured), head of cyber risk services at Travelers, said ransomware actors are already using AI to work more efficiently. In some cases, criminals who enter a company’s systems can also take advantage of the generative AI tools available within that environment.
“What we have seen ransomware threat actors do is, once they’re inside an environment, they sometimes live off the land,” Winchester told Insurance Business. “They look for what tools are available to them with the credentials they have. So, if there’s available GenAI within the environment, they might use it to make their lives easier.”
That does not necessarily make an AI platform the initial entry point for an attack. However, it can give an intruder another way to navigate the environment, process stolen information or accelerate later stages of an incident.
The exposure is becoming more relevant as businesses of every size incorporate third-party AI applications and internally developed large language models into everyday operations. Brokers may know that a client uses AI without having a clear view of which employees can access it, what data is being entered or whether the technology is connected to other systems.
“The only other thing is not to forget the inside risk of AI usage,” Winchester said. “Companies have been rapidly deploying AI, whether it’s third-party AI tools or their own LLMs, in every size of company. There’s risk involved with that if you’re not doing it thoughtfully and you’re not establishing AI governance.”
Cyber renewal discussions have traditionally concentrated on controls such as multifactor authentication, endpoint detection, backups and patching. Winchester said brokers should now also ask clients how AI is being deployed, the business purposes it serves and what work has been done to identify the associated risks. “I would definitely talking to my clients about: How are you deploying AI within your environment? What are your use cases, and what have you done to identify risk while you’re doing that?” Winchester said.
Those questions could uncover exposures extending beyond an approved enterprise platform. Employees may use public AI tools without authorization, upload confidential material or connect applications to company data without adequate oversight. Internally developed models may also introduce different risks depending on their permissions and the information they can retrieve.
Criminals can use AI after stealing data as well. Ransomware groups frequently exfiltrate large volumes of information before demanding payment, creating additional leverage through the threat of publication.
Previously, attackers might have known only the volume of information taken; today, Winchester said, criminals are becoming more specific about its contents. “Now you’re starting to see threat actors actually say, ‘We have 10 terabytes of data, and we know we have X, Y and Z sensitive data,’” she said. “That suggests some AI usage to comb through that data and really quickly understand what they may have obtained.”
The FBI received more than 22,000 complaints classified as AI-related during 2025, carrying reported losses of approximately $893 million. Business email compromise alone generated more than $3 billion in reported losses. Its Internet Crime Complaint Center also recorded more than 3,600 ransomware complaints, although the reported loss figure excludes business interruption, remediation expenses and other substantial costs.
AI is also allowing criminals to automate familiar attack methods. Generative AI is increasingly used to write software and scripts, support ransomware negotiations and search compromised email inboxes for transactions that attackers can redirect.
And while the defensive advice remains recognizable, the acceptable response time is shrinking. Winchester highlighted prioritized patching, managed detection and response, and phishing-resistant MFA as areas brokers should raise with clients. “You can’t rely on a 30-day patch timeline," she said, "you need to do it right away."
The FBI similarly recommends prioritizing patches for known exploited vulnerabilities in internet-facing systems, using endpoint detection and response tools and enabling MFA, particularly for webmail, VPNs and accounts accessing critical systems.
Brokers can also examine the risk-management support accompanying a cyber placement. Winchester said many carriers provide policyholders with access to security resources and specialists, including services that can help businesses assess their AI exposure.
“I would advise all brokers to make sure they understand what services are available from the carriers they’re placing their clients’ cyber business with,” she said. “There are a lot of free resources and experts out there that can work with their clients.”
