Katyal’s Boast of AI Role in Tariff Win Draws Swift Blowback

Milbank partner Neal Katyal raised eyebrows this week with a new TED Talk revealing his use of an AI tool to prepare to argue against President Donald Trump’s tariffs at the US Supreme Court.

Katyal, a former acting solicitor general who has presented more than 50 cases before the court, argued the case in November after winning a coinflip over another advocate, Akin’s Pratik Shah. The court ultimately ruled 6-3 in favor of the coalition of businesses challenging the global tariffs.

“I walked up to that mahogany podium, and I won,” Katyal said in the talk, released Thursday, before revealing what he described as his secret weapon in prepping for the case—a “bespoke AI system” trained on 25 years of Supreme Court papers that was able to predict not just the justices’ questions, but their eventual opinions.

It’s unusual to see an advocate like Katyal take such personal credit for a win at the Supreme court, and his remarks prompted pushback from court watchers, particularly considering the case in question involved multiple elite law firms and dozens of amicus briefs.

The backlash came quickly, with a torrent of posts on social media and critical columns published on the National Review’s website and The Volokh Conspiracy, a legal blog hosted by Reason magazine.

“The notion that it’s Katyal’s preparation with his mindfulness coach and AI that was the key difference maker, I just think people think that was a little preposterous,” said Daniel Epps, a professor at the Washington University School of Law in St. Louis.

Katyal undoubtedly put his own “blood, sweat and tears into the case,” said Xiao Wang, who directs the University of Virginia’s Supreme Court Litigation Clinic. But Wang, who has argued twice before the justices, said high court advocates are only one part of getting cases across the goal line.

“I thought it was a little strange for the quarterback here to say there is in fact an ‘I’ in team,” Wang said.

Katyal, who along with Milbank didn’t respond to multiple requests for comment, appeared cognizant about how his presentation might be perceived.

“I know how this sounds,” Katyal says in the talk, filmed last month in Vancouver. “Lawyer wins a big case, gets a fancy TED Talk invitation, talks for 14 minutes about how great he is.”

The TED Talk was not that, he assured the audience. Rather it was about the four coaches he’d brought onboard to help him prepare to better connect with the justices: a mindfulness coach who had worked with tennis great Andre Agassi, an improv coach, a meditation coach, and Harvey, the AI model he said helped him anticipate the justices’ thinking.

And although the TED Talk focused almost exclusively on his own preparations, in a social media post promoting it, Katyal credited the “best legal team in the nation” backing him up, in particular fellow Milbank partner Colleen Roh Sinzdak, as well as the Liberty Justice Center for organizing the business plaintiffs.

Tech Talk

The rapid adoption of LLMs and other AI tools have upended the legal industry, particularly as courts grapple with their penchant for hallucinating case citations. But Wang predicted Supreme Court advocates will increasingly use the technology for digital moot courts, as he himself has, if many aren’t already doing so.

Katyal’s AI “coach” was built by Harvey AI, a San Francisco-based company that makes large language models for the legal industry. Harvey says its client list includes Walmart Inc., Comcast Corp., and O’Melveny & Myers, and that it’s been used to prep for at least one other Supreme Court case this term.

A spokeswoman for Harvey said in an email it has no arrangement with Katyal to promote its services and he doesn’t hold any investment or equity stake in the company. Katyal didn’t respond to a request for comment Friday.

In the TED Talk, Katyal displayed Harvey’s predicted questions alongside excerpts from the actual oral argument transcript, saying some were “almost verbatim.” He also credited the tool with identifying an “escape route” to bring Chief Justice John Roberts to his side in the case.

“Harvey glimpsed that narrow door, I held the door open, the Chief Justice walked through it,” Katyal said.

Epps, who clerked for Justice Anthony Kennedy, echoed other critics who said the comments came across as “arrogant” and could undermine Katyal’s standing with the justices.

“I think he’s really damaged his credibility with them,” Epps said, adding Katyal sounded as though he was saying he was able to manipulate the justices and change their votes.

Still, Katyal’s presentation underscored the potential value of AI-driven prediction tools, according to Mitu Gulati, a University of Virginia law professor who studies the Supreme Court.

Advocates prize consistency from the justices, Gulati said, and the fact that an AI model could identify patterns in their questioning and reasoning was reassuring.

“It’s not that they’re outcome-driven, regardless of argument,” Gulati said.

Even if Katyal overstated the AI model’s role, Gulati said the talk ultimately highlighted the preparation and performance skills that have made Katyal a go-to lawyer for high-stakes Supreme Court litigation.

“Everybody reads everything and it’s hard to imagine oral argument changed anybody’s vote,” Gulati said. “That said, if you can change a vote, that’s why you pay the guy.”