Las aseguradoras esperan megademandas por la IA

Aseguradoras anticipan demandas multimillonarias derivadas de agentes de IA que se descontrolan. Executives from AI companies OpenAI and Anthropic may face liability for their models' actions as security failures caused by released AI agents, such as an OpenAI attack on Hugging Face, have prompted insurers and lawyers to consider potential lawsuits and compensation.

Aon insurance broker analyzed over 300 AI-related legal cases, finding insurers could be forced to pay claims under policies covering crimes, intellectual property, media liability, cyber security, tech errors, and omissions. Company executives could be sued for directors' and officers' (D&O) insurance covering executives if they face court for their decisions.

Current CEO of OpenAI, Sam Altman, is blamed for the Hugging Face incident due to lack of control in his business, according to Verisk insurance claims director Tim Rayner. Insurers and legal advisors agree executives could be held responsible for director and officer liability claims but note there are no precedents. OpenAI shareholders might sue directors for inadequate risk management if they can prove losses to the company.

Kevin Kalinich, Aon's digital risks specialist, explains the strength of a demand against an AI lab director depends on factors such as if the executive acted with reasonable business judgment in public statements about the company and its products. However, Hiscox CEO Aki Hussain believes it's too early to know how U.S. courts will address AI agent responsibility.

Besides legal actions against AI executives, insurers are preparing for potential lawsuits against AI creators, which could encompass product liability, algorithmic discrimination, privacy, and negligent homicide. Anthropic recently settled a $1.5 billion class-action lawsuit from authors accusing it of stealing their work to train AI models.

Other similar copyright infringement lawsuits against AI companies are pending. While no laws regulate online behavior yet, recent Meta's $18 billion settlement for alleged child protection failures and social media addiction could serve as a model for establishing due diligence duties in the digital space. Le Marquer suggests a swarm of autonomous AI models collaborating to launch a cyber attack, outside human control, could be compared to an oil spill, potentially allowing public claims for damages, similar to past tobacco and pharmaceutical industry lawsuits.