Artificial Intelligence

Artificial intelligence (AI) is the simulation of human intelligence by computer systems capable of learning, reasoning, pattern recognition, and content generation. Recent advances in machine learning and generative AI have accelerated adoption across every sector of the economy, enabling organizations to automate complex processes, enhance decision-making, and derive insights from large volumes of data.

For insurers, AI offers opportunities to streamline underwriting and risk assessments, enhance fraud detection, and increase operational efficiency. However, AI presents legal risks due to algorithmic biases, data privacy concerns, and cybersecurity threats. Insurers remain responsible for ensuring that AI-based decisions comply with applicable insurance laws, consumer protection requirements, and anti-discrimination standards. Insurers that implement governance and human oversight will be well-positioned to leverage AI while managing regulatory and liability risks.

Background

The capabilities and adoption of artificial intelligence (AI) have accelerated dramatically in recent years. What was once viewed as a specialized technology used primarily by researchers and large technology companies has become widely accessible to businesses, governments, and consumers. Generative AI systems can now produce text, images, audio, video, software code, and complex analytical outputs in seconds. Leading platforms such as OpenAI’s ChatGPT, Anthropic’s Claude, Google’s Gemini, Microsoft Copilot, Perplexity, and China’s DeepSeek have transformed how organizations analyze information, automate workflows, create content, and interact with customers.

AI is categorized as either “narrow” or “strong”. Narrow AI is designed to perform specific, task-defined functions using machine learning, natural language processing, artificial neural networks, and deep learning. Examples include generative AI tools like ChatGPT, AI coding assistants, machine learning fraud detection systems, and autonomous vehicles. Narrow AI can accomplish pre-determined tasks effectively but cannot replicate complex human cognition or reasoning.

By contrast, “strong” AI, or artificial general intelligence (AGI), refers to a theoretical system capable of performing generalized cognitive tasks across domains without human direction. The unifying convention of strong AI is machine learning indistinguishable from the human mind. AGI would be expected to reason, adapt, learn from experience, and solve novel problems in a manner comparable to human intelligence. At present, however, all AI models remain forms of “narrow AI” as no existing system has crossed the threshold to strong.

Among its many potential benefits, AI can enhance decision-making and operational efficiency, strengthen risk management and customer experience, and reduce costs. According to McKinsey & Company, 64% of survey respondents across 105 countries and industries believe AI is enabling innovation within their organizations. In healthcare, AI is applied to diagnostic and imaging analysis, personalized medicine, and clinical documentation. In legal services, AI-powered tools assist with document review, legal research, and contract drafting. Similarly, vibe-coding lowers the barrier of entry to allow non-technical developers to build functional applications, rather than writing code. While marketing and data analytics remain among the leading areas for AI adoption, few industries remain untouched by AI.

As AI becomes more ingrained in society, public concern regarding its impact on workforce displacement, algorithmic bias, digital privacy, cybersecurity threats, misinformation, and national security has heightened. Industry leaders have expressed concern that advances in AI may outpace existing legal and regulatory frameworks, precipitating an uneuqal balance between innovation and oversight. Recent attention to frontier-models, the most advanced AI engine under development, has prompted debate on how to best manage the risks associated with increasingly sophisticated technology. The rise of generative AI has further proliferated Deepfake – a form of machine learning that creates hyper-realistic images, videos, and audio. The content is difficult to distinguish from authentic media and amplifies the spread of misinformation and fraud. For example, cybercriminals use AI-generated voice cloning to impersonate family members in efforts to obtain money or sensitive information. Similarly, a popular Deepfake video mimicked Mark Zuckerberg, admitting that FaceBook’s design and goal was to manipulate and exploit its users. Governments across the world are evaluating regulatory frameworks intended to promote innovation while mitigating potential harms.

Injuries and Damages

As AI becomes more integrated into commercial, governmental, and consumer activities, it presents risks across numerous sectors, including employment, transportation, healthcare, education, insurance, and national security. Potential harms include algorithmic bias and discrimination, privacy violations, cybersecurity incidents, professional errors, workforce displacement, autonomous vehicle accidents, misinformation, and misuse of AI-enabled technologies.

Many of the most significant risks associated with AI arise from its rapid development and adoption. Concerns regarding transparency, accountability, data privacy, intellectual property, and the use of AI in high-consequence decisions have prompted increased scrutiny from regulators, courts, businesses, and the public.

Google and Anthropic
Beginning in 2017, the Department of Defense launched Project Maven, an initiative designed to integrate AI capabilities into military operations. The Maven Smart System (MSS) allows analysts to process substantial amounts of images and videos to identify objects and activities at a scale beyond human capability. While AI is deeply integrated into modern warfare for intelligence and strategic operations, large language models have yet to command drones or fire weapons.

Ethical concerns regarding the marriage of AI and national security have long been debated. In 2018, Google declined to renew its contract with the Pentagon following concerns that its AI technology would be applied to deadly military operations. Over 4,000 employees signed a petition objecting to the company’s development of warfare technology designed to accelerate image recognition and aid drone strikes, arguing that it rendered Google complicit in lethal warfare. Following its withdrawal, Google adopted new AI principles that “promote innovation and further [its] mission to organize the world’s information and make it universally accessible and useful.” Google committed against producing AI applications likely to cause harm, be used as weapons, cause injury, or facilitate unlawful surveillance.

In 2026, Anthropic publicly disputed the Department of Defense’s position on how its AI systems could be used in national security operations. The private company had secured a $200 million contract to provide advanced AI capabilities to support defense initiatives; however, its usage policies prohibited its technology from inciting violence or being used in the development of weapons. After the Pentagon requested that Anthropic remove certain safeguards and provide “full, unrestricted access to Anthropic’s models”, Anthropic declined the request. CEO Dario Amodei stated that while autonomous weapon systems “may prove critical for our national defense, frontier AI systems are simply not reliable enough to power fully autonomous weapons.” In turn, the Trump administration ceased the use of all Anthropic systems and designated the company a supply-chain risk to national security. The dispute illustrates the debate surrounding AI’s role in defense operations and the degree of oversight necessary to balance security, innovation, and global competitiveness.

Product Liability
Traditionally, product liability principles assume that a product’s design, manufacturing, and warnings are created by human actors. AI complicates this framework by becoming involved in product design, engineering, software development, and operations. Consequently, courts are questioning whether consumer-facing AI applications should be treated as products rather than services and whether resulting harms arise from design defects, inadequate warnings, or foreseeable misuse. According to K&L Gates, as AI functionality becomes embedded into everyday use, plaintiffs are more likely to describe AI experiences as a product and target the architecture of the system, such as guardrails, marketing, and defaults. Additionally, unlike physical goods, AI systems evolve through software updates and model retraining, complicating the allocation of liability as the system iterates. The European Product Liability Directive (PLD) holds manufacturers liable for software updates and cybersecurity breaches that compromise product safety. As AI becomes embedded in infrastructure, vehicles, consumer products, and medical devices, it will be imperative that insurers and underwriters evaluate the governance, product testing, and human oversight of the AI-backed product.

Accident Liability
The expansion of autonomous driving cars has raised questions about accident liability. Waymo, the first full autonomous ride-hailing service, operates in 11 U.S. cities and has driven over 170.7M customers as of 2026. Proponents argue that autonomous cars are involved in fewer crashes and relieve users' attention to other tasks. Driving requires continuous attention and decision making, and AI systems can reduce human error, which can contribute to motor vehicle accidents. For example, Waymo reports 92% fewer serious injury crashes compared to the average human driver. According to Goldman Sachs, the number of autonomous robotaxis are projected to reach 35,000 in 2030 – illustrating the forecasted scale of autonomous vehicles in the U.S.

However, autonomous systems remain susceptible to sensor limitations and software failures, especially in unfamiliar environments and conditions. In the event of a crash, determining liability is complicated and will depend on the level of vehicle automation. The Society of Automotive Engineers (SAE) classifies vehicle autonomy on a scale from Level 0 (no automation) to Level 5 (full automation). At Levels 0 through 2, the human driver remains responsible for controlling the vehicle and operates under traditional liability rules. Beginning at Level 3 automation, responsibility shifts from the driver to the manufacturer, software developer, or system provider. These vehicles can drive autonomously but require human attention and intervention when necessary. Level 4 and 5 vehicles, such as Waymo and certain Tesla systems, operate with minimal human involvement, offloading the liability from the driver. As autonomous vehicles become more common, it is expected that regulations for manufacturing standards, passenger responsibilities, and liability for damages will evolve and differ between states.

Environment
The rapid development and deployment of AI comes with environmental consequences, like increased electricity demand and water consumption, that are hard to mitigate. The computational power required to train and continuously operate AI systems consumes staggering amounts of electricity, increasing carbon dioxide emissions and pressures on the electrical grid. And, as adoption continues to grow, so does the construction of large, temperature-controlled data centers that house all computing infrastructure. Scientists from Google and the University of California at Berkeley estimate the training process alone consumes 1,287 megawatt-hours of electricity, generating about 552 tons of carbon dioxide and enough to power about 120 average U.S. homes for a year. Similarly, a single ChatGPT query consumes about five times more electricity than a simple Google search. Unfortunately, the convenience of AI and limited public knowledge about the environmental impacts of AI are only inciting more usage.

In addition to energy demands, water is required to cool hardware used for training and deploying AI systems. For every kilowatt-hour of energy consumed by a data center, it requires two liters of water for cooling to absorb the heat produced by the equipment. Increased water consumption places additional strain on local water supplies and ecosystems, particularly in communities already facing resource constraints. Many of the 1,700+ data centers in the U.S. are in rural communities with grids predominantly fueled by fossil fuels that released greenhouse gases when burned. The development of additional data centers may alter agricultural land-use patterns, contribute to biodiversity loss, and require policymakers to carefully balance the economic benefits of AI development against its environmental and societal costs.

It is important to note that AI expands our capabilities to model, measure, and manage ecological challenges. Algorithms that track extreme weather and model wildfire patterns can process data significantly faster than humans. Similarly, AI can identify trends that lead to environmental benefits more efficiently than humans. According to researchers, the development of more sustainable AI products and systems will require market-wide reforms, such as low-carbon data center infrastructure, transparent and accessible environmental reporting, and the utilization of responsibly sourced and energy-efficient computing hardware.

Legislation and Regulation

The precipitous rise of AI has left regulators and oversight committees scrambling to establish appropriate regulatory and compliance frameworks.

United States
The United States employs an innovative and sector-specific approach to AI governance. In 2025, President Trump issued Executive Order 14179, Removing Barriers to American Leadership in Artificial Intelligence, to accelerate AI innovation and promote “human flourishing, economic competitiveness, and national security”. The core tenet of the order is to ensure AI is free from “ideological bias or engineered social agendas”. It directed federal agencies to review and revise policies that could impede AI development and emphasized maintaining U.S. leadership in the increasingly competitive global AI market. Additionally, the order revoked previous AI policies that obstructed the United States’ position on American AI innovation.

The White House subsequently released America’s AI Action Plan, accompanied by three AI-related Executive Orders (EO 14318, 14319, 14320), that prioritized deregulation, national competitiveness, and rapid adoption. The Action Plan restricts federal procurement of AI systems incorporating DEI-related ideologies and advances the global export of American AI technology while reducing international dependence on systems developed by foreign adversaries, thereby enhancing U.S. leadership in the field. Additionally, it streamlines permitting to expedite the development of data centers on federal land and develop new security standards that protect against cyberattacks.

On June 2, 2026, President Trump signed Executive Order 14409, Promoting Advanced Artificial Intelligence Innovation and Security, signaling a more targeted regulatory approach to frontier AI models. Under the new order, AI companies are asked to voluntarily give the government up to 30 days to review their new AI models before releasing them to the public. The process offers the federal government oversight into AI models to identify potential vulnerabilities and mitigate risks by foreign adversaries or bad actors. It also directs the Treasury secretary to develop an AI “cybersecurity clearinghouse” to assess national security vulnerabilities associated with AI. Immediately following its release, top executives at Microsoft, OpenAI, and Google praised the order as “an important step” to balance AI safety and innovation, while other expressed concern that the order could slow development and stiffen regulation in the future.

Additionally, the National Institute of Standards and Technology (NIST) offers an important benchmark for responsible AI governance by addressing risks associated with AI. The AI Risk Management Framework (RMF) is a “voluntary, rights-preserving, non-sector-specific, use-case agnostic” guide for companies involved in the development of AI systems to manage the risks posed by the technology. The RMF is intended to provide a way to operationalize the principles outlined in the AI Bill of Rights across a range of organizations.

However, comprehensive federal privacy and AI legislation remains absent. The American Data Privacy and Protection Act failed to pass in 2021, which would have embedded strict civil rights and algorithmic accountability standards into AI regulation. Consequently, governmental oversight proceeds through state and local governments, in addition to statutory regimes, like consumer protection laws, civil rights laws, financial services regulation, employment laws, and sector-specific agency enforcement. As of June 2026, all 50 states have introduced AI-related legislation, and 38 states have adopted or enacted around 100 measures in a single year to regulate or study the technology. For example, New York City prohibits employers from employing AI-assisted hiring tools unless the product underwent independent bias audits and public notice (Local Law 144). Colorado requires developers and deployers of Automated Decision-Making Technology (ADMT) to inform users of its utilization and facilitate human review to prevent algorithmic discrimination (SB 26-189). Importantly, insurers subjected to Colorado’s existing algorithmic discrimination insurance statutes are exempt from certain obligations that govern underwriting.

Across all states, an insurer that uses an internally developed or third-party AI system remains responsible for ensuring that the system complies with applicable insurance laws, unfair trade practice statutes, privacy obligations, and anti-discrimination requirements. The NAIC Model Bulletin on the Use of Artificial Intelligence Systems by Insurers guides the regulation for evaluating insurers’ use of AI in underwriting, pricing, and claims handling, among other insurance functions.

While each state’s regulatory approach is different, a few central principles have emerged. The federal government is prioritizing innovation, national security, and global competitiveness, while state regulators are increasingly focused on transparency, bias, and consumer protection. For organizations, AI might be used to improve efficiency and risk assessment; however, it must be supported by governance and human oversight that protects users from discrimination or unlawful claims. Likewise, improved federal transparency is widely agreed upon. In September 2025, the Pew Research Center reported that six in ten Americans want greater control over the role of AI in their lives. Many lawmakers believe that when algorithms are used to inform consequential decisions, public disclosure of their use is essential for effective governance and trust. As such, a growing number of states require registration of AI systems and efforts to mitigate system bias.

Europe
For companies operating in Europe, the AI landscape is governed by the European Union Artificial Intelligence Act (EU AI Act) which launched in August 2024 and is being implemented in phases through 2027. As the first major regulator of AI, the legislation strengthens rules around data safety, ethics, and innovation by subjecting the providers and deployers to specific legal obligations. The Act adopts a risk-based approach to regulating AI, whereby the requirements for a system are relative to the level of risk it poses to the “health and safety or fundamental rights of a person.” First, applications that establish an unacceptable risk, such as social scoring systems and manipulative AI, are highly regulated and banned. Secondly, “high-risk” AI systems that pose significant threats to health, safety, or fundamental rights, like CV-scanning tools that rank job applications, are subject to rigorous regulation and testing. Additional examples of “high risk” AI include autonomous vehicles, medical devices, and critical infrastructure machinery. Under the EU AI Act, the use of AI in health and life insurance for risk assessment and pricing is also considered a high-risk activity.

“Limited risk” AI systems that interact directly with humans or generate synthetic content (chatbots) are governed by lighter transparency obligations to ensure end-users are aware that they are interacting with AI. Developers of these models are required to “apply safety checks, data governance measures and risk mitigations” before being released to the public. In addition, providers must ensure that the data used to train their systems complies with copyright and intellectual property laws. Lastly, many AI systems we interact with daily pose minimal risk to human safety and are coined “minimal risk”. They are largely unregulated and carry no extra obligations.

The following list of regulations provides additional information related to the current legislative landscape for artificial intelligence.

The Unfair Claims Settlement Practices Act
The Unfair Claims Settlement Practices Act (UCSPA) outlines minimum standards for states to regulate insurance carriers and protect policyholders from “bad faith” behavior by insurers during the claims process. While the Act predates modern AI technologies, it requires insurers to conduct reasonable investigations, communicate promptly with claimants, and base claim decisions on factual grounds – obligations that apply equally to AI-assisted decisions, too. If insurers utilize machine learning and generative AI tools to evaluate claims, detect fraud, or recommend coverage decisions, they must mitigate the risks of AI hallucinations and “black box” decision-making. Ultimate responsibility for claim decisions remains with the insurer regardless of the technology employed. Accordingly, human oversight and validation procedures are necessary when incorporating AI into the claims handling process.

NAIC Model Bulletin on the Use of Artificial Intelligence Systems by Insurers
The NAIC Model Bulletin is the leading U.S. insurance-specific AI governance framework. It directs insurers to maintain governance, risk-management, testing, monitoring, documentation, and vendor-oversight controls for AI systems used in underwriting, pricing, claims handling, fraud detection, marketing, and customer service.

NAIC AI Systems Evaluation Tool Pilot
A 12-state initiative to inform regulators on how domestic insurers use AI and assess whether their governance practices are effective in managing the risks. It empowers states to review AI systems, promote transparency, and identify areas of oversight or improvements. The pilot program will determine whether the tool helps insurers explain their AI governance systems to regulators, how regulators receive those standard governance practices, create long-term recommendations for market conduct and financial risk assessment review processes, and identify what addition regulator training may be needed in the future. The pilot program will run through September 2026.

National Artificial Intelligence Initiative Act of 2020 (“NAIIA”)
The NAIIA establishes statutory guidance for national AI development, accountability, and highlights the need for open data and best sharing practices to be incorporated into AI standards. In addition to risk management frameworks, the act authorized the National Science Foundation to fund and oversee AI research on its applications and ethical challenges. It also established the National Artificial Intelligence Advisory Committee, an interdisciplinary group of experts from academia, society, and the private sector to provide recommendations on the current state of U.S. AI innovation.

Colorado Artificial Intelligence Act (SB 24-205)
Colorado’s AI Act is the first broad U.S. state statute regulating high-risk AI systems. It requires developers and deployers to use reasonable care to prevent algorithmic discrimination and impose governance, disclosure, documentation, and impact-assessment obligations.

New York Department of Financial Services Circular Letter No. 7
NYDFS Circular Letter No. 7 governs the use of AI systems in insurance underwriting and pricing. It mandates anti-discrimination testing, actuarial validity, governance, and third-party oversight.

NHTSA Standing General Order on Crash Reporting for Automated Driving Systems and Level 2 Advanced Driver Assistance Systems
The Standing General Order requires manufacturers and operators of vehicles equipped with automated driving systems or Level 2 advanced driver-assistance systems to report certain crashes to the federal government. It is the principal federal reporting framework for autonomous and semi-autonomous vehicle safety data.

Illinois Biometric Information Privacy Act (“BIPA”)
BIPA regulates how private entities collect, store, and use biometric data. It requires companies to obtain written consent, inform individuals of what data is collected, and release policies on retention and destruction of data. With AI advancements utilizing facial recognition, identity verification, and monitorization, the protection by BIPA ensures individuals have control of their own biometric data.

Department of Defense Directive 3000.09 – Autonomy in Weapon Systems
DoD Directive 3000.09 dictates the U.S. Department of Defense policy on autonomous and semi-autonomous weapons. The directive ensures operators exercise human judgement over force and establishes the safety guidelines to minimize unintended accidents. The directive requires that all systems allow human use, undergo rigorous testing, and undergo frequent reviews before they are deployed by the U.S. military.

California AI Transparency Act (SB 942)
This consumer protection law combats the spread of generative AI by requiring developers to embed watermarks into all AI-generated content. This no-cost tool ensures users can verify whether content was altered by AI and reduces consumer deception and fraud. The statue is significant for synthetic media and Deepfake content.

TAKE IT DOWN Act
The federal TAKE IT DOWN Act criminalizes the nonconsensual sharing of intimate images developed by AI software and requires online platforms to remove reported content. Not only does the Act protect victims, but it also encourages websites to establish clear policies and removal processes to retain a safe digital space.

Liability and Insurance

AI is being adopted across industries at a rapid rate. In a study conducted by the NAIC in 2025, 88% of personal automobile insurers and 70% of homeowner insurers are “currently using, planning to use, or exploring the use of AI”. The technology is being incorporated into a variety of insurance operations, including product development, ratemaking, claims payments, customer relations, fraud detection, distribution, and underwriting. And, it has the opportunity to create lasting business value if scaled thoughtfully.

For most effective use, experts recommend that insurers embed the technology across their organization, rethink current operating models, and rewire how AI interacts with various domains, such as underwriting, claims, and customer service. AI is more valuable when scaled across the business, even if the back-end results are different. For example, customer service requests can be repurposed to develop IT support, marketing strategies, FAQs, or legal drafting. As AI continues to develop, the insurer industry can expect AI to be capable of all onboarding functions, such as: • Intake Agent: Collects and organizes customer information, including data contained in complex records and reports. • Risk Profiling Agent: Builds risk profiles, abiding to existing underwriting standards. • Product Agent: Price the case and develop policies that meet customer needs. • Compliance Agent: Review product to ensure integrity of regulatory and ethical standards. • Managing Agent: Determines whether an application may be processed automatically or should be escalated for human review. • Human Oversight: Provides supervision and validation to mitigate algorithmic bias and ensure regulatory compliance. However, while AI offers immense promise, few insurers have fully integrated it into their business models. The NAIC outlined the mounting concerns regarding appropriate use of AI in the industry that impede broad adoption.

First, AI systems are only as reliable as the data it is trained on and may incorporate biases that influence insurers' decision making. Such biases may result in unfair discrimination of insureds and impact their eligibility, terms of coverage, and price they pay. For example, the class action lawsuit, Huskey v. State Farm Fire and Casualty Company, alleged that State Farm’s use of AI fraud detection disproportionally flagged Black policyholders. In the health insurance sector, the use of AI to handle claims could instigate litigation or expose sensitive data in the event of a data breach. Similarly, using AI for prior authorization might prevent patients from receiving medical care if they are rejected. For example, the AI-based medical insurance company, HealthCare Interactive, experienced a data breach in 2025 that exposed patient names, lab results, and health insurance enrollment data, among other items.

Following President Trump’s Executive Order that reduced national and state involvement in AI regulation (EO 14179), the NAIC expressed concern that it could undermine the long-established system of state insurance regulation and create uncertainty regarding regulator’s ability to oversee insurers’ use of AI. The NAIC emphasized that state insurance regulators have historically been responsible for ensuring that underwriting, rating, and claims practices remain fair, transparent, and free from unfair discrimination. However, broad federal preemption could impair regulators' ability to address risks associated with AI decision making in critical insurance functions, including rate setting, underwriting, fraud detection, and claims processing. The NAIC further cautioned that regulatory uncertainty may delay business decisions, discourage investment, and postpone consumer protections at a time when AI adoption across the insurance sector continues to accelerate. Regardless, insurers are expected to uphold robust AI governance and remain accountable for the outputs of AI systems used throughout the insurance lifecycle.

Employment Practices Liability Insurance (EPLI)

As AI continues to become implemented into the workplace, Employment Practices Liability Insurance is likely to experience a corresponding rise in claims involving automated employment practices. Employers are frequently using AI systems to improve efficiency in the hiring process, such as resume screening, candidate ranking, and interview analysis; however, they also risk algorithmic bias and discriminatory outcomes. For example, Amazon discontinued an experimental AI recruiting tool after it systemically disadvantaged female applications because it had been trained on a predominately male dataset. Similarly, employers often collect and process employee information through AI systems. Improper collection or usage of such biometric and personal data may result in claims arising under employment or privacy statutes.
Companies are also deploying AI to streamline operations, reduce costs, and eliminate routine jobs. According to Challenger, Gray & Christmas, U.S. employers dismissed more than 108,000 jobs in January 2026, marking the worst month for layovers since 2009. As such, economic pressures, widespread adoption of AI, and varying state laws are likely to further increase ELP claims. Munich Re suggests that claims alleging wage-and-hour violations and lack of accommodation under labor laws are likely to rise due to the confluence of AI, hybrid work trends, and rising labor costs.

Commercial and Personal Property Insurance

AI serves as a valuable tool to support the commercial and personal property insurance industries. In the commercial market, AI can analyze large volumes of loss, weather, and operational data to generate risk models and simulate potential loss scenarios. Personal property and casualty insurers can use AI to automate claims processing, analyze property damage, and improve fraud detection. For auto insurers, AI can assess driving records and demographic analyses of customers to estimate the risk involved in covering a driver. As mentioned, while AI improves efficiency and transforms risk assessments, insurers must engage in human review to ensure outputs are valid and comply with insurance laws.

Health Insurance

AI can analyze large data sets to help predict patient outcomes and develop more personalized health plans. However, concerns that AI-backed insurers will deny health coverage and preclude at-risk individuals from receiving necessary care is a hurdle insurers must consider.

Litigation

AI is reshaping industries across the economy and, in turn, generating demand for legal services. Leading topics of legislation include harms caused by deepfakes, disruptions in the workplace, algorithmic pricing and housing costs, development of new data centers and energy infrastructure, and impacts on data privacy. The following cases provide insight into the current legal environment.

The New York Times Co. v. OpenAI, Inc. and Microsoft Corp.
The New York Times filed suit against OpenAI and Microsoft, alleging that millions of copyrighted news articles were used to train large language models without authorization. The Times contends that the defendants’ AI systems generate outputs that closely summarize copywritten content, thereby competing with its journalism and subscription business. The litigation has become the precedent case addressing whether copyrighted material used for AI training constitutes fair use under U.S. copyright law. The outcome will influence future licensing arrangements, data training practices, and intellectual property rights in the AI industry.

Mobley v. Workday, Inc.
Mobley filed a class action lawsuit alleging that Workday’s AI applicant screening tool discriminated against race, age, and disability. The plaintiff is an African American male over the age of 40 who suffers from anxiety and depression. Despite holding a bachelor’s degree and applying to roughly 100 jobs through Workday, Mobley was consistently denied employment. The complaint alleges the AI tools provided by Workday allow employers to use “discriminatory and subjective judgments” when evaluating applicants. The litigation is one of the first major cases testing whether companies are liable for discriminatory outcomes produced by automated hiring systems. It also signals that employers cannot avoid liability by relying on third-party AI tools. The outcome of this case will inform liability standards towards algorithmic bias and employer accountability.

Getty Images (US), Inc. v. Stability AI Ltd.
Getty Images alleges that Stability AI copied millions of copyrighted photographs without authorization to train its image-generation models. The case represents another copyright dispute involving generative AI and will inform whether AI developers may use protected creative works as training data without obtaining licenses.

Clearview AI, Inc. Consumer Privacy Litigation
Clearview AI has faced numerous lawsuits alleging the use of billions of images from social media and public websites without consent for the development of facial recognition technology. The litigation concerns violations of biometric privacy laws, including the Biometric Information Privacy Act (BIPA). The case presents the legal risks associated with the collection, storage, and commercial use of biometric data for AI applications.

Doe v. Character Technologies, Inc.
Wrongful death and product liability lawsuits filed against Character.AI allege that AI interactions contributed to psychological harm of minors. Plaintiffs contend that the company failed to implement adequate safeguards and encouraged harmful behavior. In 2025, a federal judge dismissed Character.AI’s motion to dismiss, indicating that AI chatbot’s outputs are not protected under free-speech rights. In turn, these cases represent the safety of AI, and the duties owed by AI developers to users.

Tesla Autopilot and Full Self-Driving Litigation
Tesla faces litigation from accidents involving its Autopilot and Full Self-Driving technologies. Plaintiffs believe that Tesla overstated the capabilities of its driver-assistance systems and failed to adequately warn users of its operational limitations. These cases have the power to inform the future of autonomous systems, consumer disclosures, and manufacturer responsibility in the U.S.

State of Florida v. OpenAI, Inc.
In June 2026, Florida became the first U.S. state to file suit directly against OpenAI, indicating that ChatGPT was released without adequate safeguards and contributed to harmful conduct, including self-harm, violence, and dangerous advice. While the case is in its initial phase, the litigation reflects a growing trend toward AI safety and product-liability claims targeting developers of frontier AI model.

Future Outlook

AI is expected to become increasingly integrated throughout the insurance lifecycle, from underwriting and claims handling to fraud detection, customer service, and product development. As technology advances, insurers will likely deploy more sophisticated AI systems capable of automating routine functions, improving risk modeling, and delivering personalized insurance products. At the same time, evolving regulatory frameworks and increased litigation may require insurers to strengthen governance, transparency, and human oversight. The organizations best positioned for long-term success will be those that leverage AI to enhance operational performance while maintaining compliance, consumer trust, and accountability.