Anthropic lawyers predict AI automation for junior attorney work

Anthropic’s general counsel offered blunt advice to Stanford Law School students earlier this month: law schools are failing to prepare them for a profession artificial intelligence (AI) is already transforming.

“Law schools are doing a terrible job,” Ethan Forrest, head of Product Legal at Anthropic, told a room of roughly fifty Stanford law students at “Governing Frontier AI: A Conversation with Anthropic’s General Counsel,” an event hosted by the Stanford Law and Technology Association on Jan. 15. He said law students shouldn’t be taking tests or be enrolled in traditional classes, arguing that AI now performs most tasks assigned to first- and second-year associates “better than most junior lawyers.”

Forrest’s assessment comes as Anthropic, the AI company behind Claude, prepares to sell its stock for the first time this year, according to a New York Times story report published the day before the event. During the discussion, Anthropic’s deputy general counsel Aparna Sridhar acknowledged the company’s precarious position in a fiercely competitive industry. “We really believe that there is some chance that [Forrest and I] will both be looking for new jobs this year,” she said.

The speakers said Claude already saves Anthropic’s legal team one hour per lawyer per day by automating tasks like tracking product launches and analyzing spreadsheets, work traditionally delegated to junior associates. “All the work that first- and second-year attorneys tend to do at firms where you are being directed by a more senior lawyer to give them structured inputs basically, that’s a machine’s job now,” Forrest said.

Forrest placed responsibility for adapting to AI squarely on law firms and law schools. He criticized law partners for failing to train young lawyers in fundamental skills like arguing, negotiating and writing and instead burning them out with repetitive tasks. “Nothing I have done in the last 20 years I have used, including all I have learned from law school,” he said. 

Sridhar encouraged students to leverage their familiarity with AI tools as an asset, as they will be “the first set of lawyers that are truly AI native.” 

“You are going to know a lot more about the internal AI tools and the legal AI tools that are in the market than a lot of the partners that you may be working with,” Sridhar said.

The executives also defended Anthropic’s legal strategy as it navigates copyright challenges. The company recently settled the Bartz v. Anthropic case, in which authors sued over the use of copyrighted books to train Claude. Forrest explained that while Anthropic believed its fair use argument was strong — that training AI on books creates an AI model fundamentally different from the original works — the potential damages were too risky to litigate.

“There are statutory [damages] to the tune of up to $150,000 per work, and these datasets have millions of books in them, so the math quickly gets uncomfortable,” Forrest said. “Sometimes, settlement is the better part of our work.”

Both speakers expressed pessimism about U.S. legislative action to regulate AI. Forrest noted that Congress does not have capacity to pass AI laws and that the Trump administration has taken a hands-off approach in AI regulation. By contrast, both said Anthropic’s lawyers now spend as much time on international regulations like the European Union’s AI Act as they do on U.S. law, a reality that raised questions about whether American law schools are preparing students for an increasingly global technological and legal landscape.

They agreed that half their work involves “being a middle manager, not a lawyer,” as they frequently address issues like copyright disputes, mental health concerns and AI safety regulations across multiple jurisdictions.

Stanford law student Adèle Boreau-Potocki said the talk provided an “eye-opening” perspective on being an AI-native lawyer, and that the speakers’ advice to use AI as a professional asset was valuable for career planning.

“We can own [fluency in AI] and make the most out of our skills,” Boreau-Potocki said. “It can be a differentiating factor for being hired, knowing how to use AI tools.”

Forrest concluded with a stark vision of modern legal practice: clients now expect only four-bullet-point summaries of litigation and memos. “You put in a fifth bullet, you are rolling the dice,” he said.

The post Anthropic lawyers predict AI automation for junior attorney work appeared first on The Stanford Daily.