Big Law Ramps Up AI Training to Combat Breach Threats, Misuses

A majority of law firms are offering—and some even requiring—generative AI training for attorneys, in part to help stave off risks of data breaches and unauthorized use of the technology.

About 70% of law firms that reported on technology training in Bloomberg Law’s 2026 Leading Law Firms survey said they offer generative AI training for attorneys. Among those who offer such training, about 73% of attorneys on average have completed it.

Firms said training, and having governance teams, helps attorneys learn and experiment with AI securely, reducing the risk of leaking private and confidential data that leads to regulatory violations and made up citations.

About 57% of law firms that participated in the survey—which was self-reported—said they have AI governance teams.

“If you’re not giving your attorneys and your staff a option for using AI safely and responsibly, they’re going to find it on their own and they’re going to experiment with their own,” said Robert Skinner, the CEO at Offit Kurman, a mid-size firm.

Training Benefits

Firms looking to leverage AI are taking steps to encourage their attorneys to get familiar with technology and to act responsibly.

For example, Eversheds Sutherland has been urging employees “to spend at least two hours per week of their own personal time to become familiar with their favorite AI tools,” according to  Lino Mendiola, global co-CEO and US chief executive of Eversheds, which reported high levels of technology use in the survey.

“Again, that’s on top of everything that the firm asks of them. And that was the call to action,” Mendiola said.

Read More: See which firms came out on top overall in Bloomberg Law’s Leading Law Firms.

That move was also paired with training, as well as embedding AI thought leadership, he said.

Offit Kurman offers AI training to make sure their attorneys have clear rules and a safe path to integrating AI in their workflow.

The firm, for example, uses multiple layers of training, through its professional development team on AI tools. Attorneys are trained on how to spot and safely manage the output of their work through instructors, as well as computer-based training. Offit Kurman reported that all of its attorneys have received AI training.

“We think that the training is required not only for meeting our duties of responsibility as a law firm, but also making sure that through the governance and training, we’re making sure that we’re not discouraging our attorneys from using AI. We want them to experiment responsibly and safely,” Skinner said.

Minimizing Risk

One of the biggest dangers of using AI tools without authorization is that confidential or sensitive information is entered into a public facing tool, said Michael Bruckner, the chief information officer at Duane Morris. Providing governance controls minimizes the risk of attorneys turning to those unauthorized tools, Bruckner said.

“Our philosophy on this, really, is we want to make tools available that we have the governance controls in place for, and make sure that people have access, understand how to use them, and those are the tools that they’re going to, as opposed to trying to use publicly available tools,” he said. “And we do have some controls and monitoring in place to prevent access to publicly available tools.”

Defining a list of permissible tools and their appropriate usage is also important, said Eric Felsberg, a principal at Jackson Lewis who heads the firms technology industry group.

“Not only does it make sure that people comply with whatever the guidance is, but it also gives them an idea that, ‘Hey, I now know that I can’t just go out and download any platform to start using it without any sort of protections,’” Felsberg said.

Getting Onboard

Large firms with more than 500 attorneys are more likely to offer AI training, according to the survey reporting. The largest ones, with more than 1,000 attorneys, reported a higher attorney completion percentage for those programs.

Firms that don’t set up such programs or don’t have governance teams run the risk of compromising themselves and their clients.

“If people are not at the point of having a policy, I think they’re making a bet that AI is not a transformational shift or a paradigm shift, or maybe underestimating the mental model for how widespread the technology is,” said Mark Williams, a law professor at Vanderbilt University and the co-founder of the university’s AI law lab. “There are sort of risk points and entry points for AI to be inadvertently used and abused across all of knowledge work just because of the way systems are now designed.”

That risk is ever present, even if there are policies around AI use.

“I say the future is here, but it’s unevenly distributed. When I do a lot of work with smaller to mid-size firms, they might even know that they have to do something or starting to think about it, but it’s just so overwhelming,” Williams said. “It can be a little bit of paralysis by analysis.”