Published: New England Biz Law Update
Artificial intelligence is rapidly becoming part of everyday business operations. Companies increasingly rely on AI-powered tools to screen job applicants, evaluate creditworthiness, interact with customers, and automate decisions that previously required human judgment. These technologies promise efficiency and cost savings, but they also introduce significant legal risks.
As businesses integrate AI into their operations, litigators will be increasingly faced with a novel question: when an AI system used by a company produces an inaccurate, discriminatory, or otherwise harmful result, who is liable, and for what?
The answer may turn out to be less novel than the technology. Although AI will certainly prompt new factual and evidentiary questions, existing Massachusetts and federal law already supplies the causes of action that are likely to apply, and courts have shown little patience for the argument that the machine, not the business, is to blame.
For example, in April 2024, Massachusetts Attorney General Andrea Campbell issued an advisory clarifying that the Commonwealth’s existing consumer protection, anti-discrimination, and data security laws apply to artificial intelligence and algorithmic decision-making systems. The advisory specifically identifies the Massachusetts Consumer Protection Act, Chapter 93A, the Massachusetts Anti-Discrimination Law, Chapter 151B, and the Data Security Law, Chapter 93H, as potential sources of liability for businesses developing, supplying, or using AI, and makes clear that conduct does not become lawful simply because an algorithm – and not a person – performed it. For Massachusetts businesses, the message is straightforward: AI adoption does not eliminate existing legal obligations. Instead, it introduces new ways to violate them.
One newer, emerging area of exposure involves discrimination claims based on automated decision-making systems used in employment and lending. In July 2025, the Massachusetts Attorney General announced a $2.5 million settlement with student loan lender Earnest Operations LLC concerning alleged violations of consumer protection and fair lending laws. According to the Attorney General, Earnest used algorithmic models to determine loan eligibility, terms, and pricing. Certain practices allegedly created risks of disparate harm to Black, Hispanic, and noncitizen applicants and borrowers. The settlement required enhanced AI governance and compliance measures, and illustrated that the company could not shield itself from liability by passing the buck to its AI tools.
Instead, the developer or provider of the AI tools at issue could serve as additional defendants in potential claims, along with their customer. For example, in Conn. Fair Hous. Ctr. v. CoreLogic Rental Prop. Sols., LLC, a federal court in Connecticut allowed Fair Housing Act claims to proceed against a tenant-screening vendor whose automated product disqualified an applicant, reasoning that the vendor could be liable both directly and as the landlord’s agent. And in Cornish-Adebiyi v. Caesars Ent., Inc., the Third Circuit revived Sherman Act claims alleging that casino hotels and their AI pricing vendor used shared software to fix room rates.
Nor should businesses expect the “AI did it” defense to protect them. Sanctioning a lawyer whose AI-assisted filings cited nonexistent cases, a Massachusetts Superior Court judge, in Dastou v. Holmes in 2025, wrote that one who uses generative technology “may not blame errors in their work product on the technology.” There is little reason to expect a more forgiving rule for businesses generally.
Plaintiffs should, and will, continue to evaluate the extent to which existing theories of liability can be applied to harm caused by AI tools. Businesses and their counsel should seek to address this added legal threat by verifying AI outputs as much as possible, instituting internal practices and policies to mitigate these concerns, and – where possible – allocating risk contractually with AI vendors because the groundbreaking nature of this technology, and its impact on business, isn’t changing any time soon.