Hidden AI Risk: What Every Tech Company Should Know About Liability and CoverageTech companies are riding high on the AI boom. According to UN Trade and Development, AI technology accounted for just 7% of the frontier tech market in 2023. By 2033, it is expected to represent 29% of the frontier tech market, with a projected value of $4.8 trillion. For tech companies, the business opportunities are huge, but so are the risks.
With any product or service, there’s a risk of software bugs, missed deadlines and other failures. AI tools have these risks, but they also have additional liability exposures unique to the nature of AI, and traditional technology insurance solutions may not fully address them.
Here are three risks that tech companies should be aware of.
AI Risk #1: Intellectual Property Liability
Multiple lawsuits have accused tech companies of violating intellectual property rights by using copyrighted material to train their AI models without permission, sometimes through the use of pirated copies. In one lawsuit, Susman Godfrey secured a $1.5 billion settlement against Anthropic to settle the company’s use of pirated databases.
In another major lawsuit, The New York Times has sued OpenAI and Microsoft over allegations of copyright infringement. According to Harvard Law Review, the lawsuit claims that OpenAI committed copyright infringement when it used news articles to train its models, and that the resulting large language model (LLM) sometimes memorizes and generates near-verbatim reproductions. The lawsuit also argues that the LLM reproduces more content than the publisher would show online with a subscription, thereby harming the publisher by allowing readers to circumvent the paywall.
What does this mean for tech companies? Tech companies need massive amounts of data to train AI, but when copyrighted material is used in training without permission, and when outputs resemble copyrighted material, allegations of copyright infringement may follow.
AI Risk #2: Privacy and Personal Data Exposure
AI has raised several concerns over privacy.
As with copyright infringement allegations, some of the issues involve the data used to train AI. According to Business Insider, a proposed class-action lawsuit accused OpenAI of secretly harvesting personal data to train its models. The data used allegedly included medical records and information about children.
AI models have also been accused of privacy violations stemming from misappropriation of likeness. According to the Authors Guild, authors have sued Grammarly over its creation of a feature called “Expert Review,” which presented writing tips as if they came from real authors. However, those authors never agreed to the feature, and some complained that the advice attributed to them was not something they would say and could harm their reputation.
Meanwhile, Grok is facing litigation over its AI image generation tool. According to AWKO Law, people can reportedly use Grok and other image-generation tools to create non-consensual intimate images of real people.
What does this mean for tech companies? Tech companies that collect, store or use personal information may face allegations of privacy violations. AI outputs can also run afoul of privacy rights, and tech companies may be held accountable for the outputs created by users.
AI Risk #3: Harmful User Interactions
AI chatbots may reinforce and amplify users’ delusions, according to Psychology Today. Cases of so-called “AI psychosis” include people who suffer from grandiose delusions and think they have uncovered some truth about the world, people who believe the AI is a deity, and people who believe the AI is a romantic partner. Some people with no history of mental illness have experienced AI psychosis leading to psychiatric hospitalizations, and one man was killed in an encounter with police.
AI chatbots may also encourage suicide, according to multiple allegations. According to CBS News, one lawsuit accused ChatGPT of encouraging a man to commit suicide. According to Reuters, another lawsuit accuses ChatGPT of encouraging a young woman to commit suicide after multiple chats involving suicidal ideations. At least 18 similar lawsuits have been filed in California.
What does this mean for tech companies? Without proper safeguards in place, users can become attached to AI in very unhealthy ways, and AI companies may face product liability and wrongful death litigation as a result.
Does Your Insurance Cover AI-Related Liability?
Tech E&O insurance is a staple of liability protection for tech companies, but it may not cover all AI-related liability exposures.
As insurers try to manage AI risks, they are reconsidering policy language. According to Carrier Management, insurers are trying to decide how AI should be covered under cyber and tech E&O policies. Depending on what they decide, they may introduce new AI exclusions.
Even without exclusions, some claims may fall outside the scope of tech E&O. For example, a standard tech E&O policy does not cover wrongful death claims. To fill in coverage gaps, tech companies may need other types of insurance coverage, such as product liability that covers AI.
Without adequate coverage, companies could face expensive defense costs, settlements, awards and regulatory penalties without the support of insurance.
Just as AI is a rapidly evolving area of tech, AI insurance is a rapidly evolving area of coverage. Heffernan Insurance Brokers can help your tech company assess its AI-related risks, identify potential coverage gaps and build an insurance strategy that supports growth and innovation. Learn more about our insurance solutions for tech companies.
