The stakes are sky-high in recently filed lawsuits that accuse Google, Meta, Microsoft and OpenAI of breaking the law when they gather training material for their large language models from public databases, called “scraping” in tech circles.

As with many of the practices surrounding artificial intelligence, experts in privacy and intellectual property say scraping falls in a legal gray area that will become better defined as courts weigh in on these and other cases, establishing guardrails for what’s allowed.