Inside the AI boom that's changing how Big Law attorneys work

Business Insider

18-07-2025

DLA Piper rolls out Microsoft Copilot firmwide
Assess: DLA Piper has defended Microsoft in a defamation suit over AI-generated content and helped OpenAI put forward its views to Congress on how AI should be regulated. It's leaning into the tech internally, too.
Danny Tobey, chair of DLA Piper's AI and data analytics practice, said the firm has an internal group of lawyers and technologists who test tools and develop metrics for quality and accuracy. The team runs A/B tests on real cases, comparing results from traditional legal teams against AI-assisted ones to evaluate performance across speed, accuracy, and cost.
Apply: Microsoft has highlighted DLA Piper as the first major law firm to adopt Copilot firmwide, after starting with several hundred licenses in late 2023.
Lawyers use Copilot within their existing Microsoft 365 apps, Tobey said. Think drafting documents, poring over spreadsheets, and creating PowerPoint slides. For more advanced legal research and analysis, he said, attorneys turn to legal-specific tools like Harvey, CoCounsel, and LexisNexis Protégé.
DLA Piper has also developed custom language models to help clients spot compliance risks early, including under laws like the Foreign Corrupt Practices Act and the Anti-Kickback Statute.
"We've found a number of issues before they metastasized into outright violations," Tobey said, "and that allowed the company to step in and do some education and compliance refreshing before there was a problem."
Align: Tobey said the firm provides detailed training for lawyers on how to use its tools.
"We train on a per-tool basis because they all have strengths and weaknesses," Tobey said. "If you were a doctor, you would not adopt a new tool without being trained in its limitations."
Gibson Dunn pilots ChatGPT Enterprise with its lawyers and staff
Assess: Before adopting any tool, Gibson Dunn runs a three-step review process, said Meredith Williams-Range, the firm's chief legal operations officer.
Tools must first pass an internal audit covering security, privacy, and risk. Next, they undergo proof-of-concept testing with a small group. Finally, tools must demonstrate real value to lawyers through hands-on use, a process that can take days or, as with a Harvey pilot, stretch over several months.
Apply: ChatGPT Enterprise is one tool making its way through Gibson Dunn's internal processes. In June, the firm launched a pilot with more than 500 participants — a mix of lawyers and staff — to put the product through its paces.
Williams-Range said she emailed practice group leaders and managing partners around the world, asking them to submit lawyers willing to test the tool. Three days later, 450 people had signed up — more than twice what she expected.
Gibson Dunn says it's also evaluating using rival AI models Google Gemini and Claude Enterprise. The firm works with a range of vendors, including Harvey, Thomson Reuters, and Microsoft. Some tools, like Harvey and CoCounsel, are used to support legal work, while Copilot helps with administrative tasks.
For more specific use cases, the firm collaborates with developers to build custom workflows tailored to its practices and data, Williams-Range said.
Align: The firm's AI policy is reviewed quarterly to stay current with changing regulations, she said. It also includes a procurement playbook with specific terms around security and how it shares learnings about the tools.
Gibson Dunn also has a strategic advisory board made up of over 30 partners across offices globally. This brain trust meets monthly to guide policy decisions, debate use cases, and determine whether tools like ChatGPT Enterprise should be limited, expanded, or customized.
"Just because we can doesn't mean we should," Williams-Range said, referring to the principle that guides the board's work.
Sidley Austin hones prompt engineering skills during associate orientation
Assess: Over her 29 years with the firm, corporate lawyer Sharon Flanagan has watched Sidley embrace new tech, but with guardrails in place.
The firm formed an AI council with members from its management committee, executive committee, and strategy team to set policies and identify use cases. Sidley typically starts with small-scale rollouts to pilot new tools before expanding.
Apply: Sidley has explored a range of AI tools, says Jane Rheem, Sidley's chief data and AI officer — from legal-specific platforms, to broader foundation models, to point solutions that help with timekeeping or narrative writing.
The firm declined to identify the AI tools it's testing, saying it doesn't want to endorse products that may not be part of its long-term strategy.
Flanagan says uptake has been organic among litigators and corporate and regulatory attorneys.
Align: Implementation is only the beginning, Rheem says. The firm tracks usage after deployment, gathering data and feedback from "superusers" — early adopters who experiment broadly and flag where tools are working (or not).
Sidley is also focused on making sure its youngest lawyers are fluent in the tools. This year, nearly 300 incoming associates participated in a generative AI hackathon as part of their orientation.
Ropes & Gray uses AI tools like Harvey and Hebbia to squeeze more hours out of the day
Assess: When Ropes & Gray finds an AI service it likes, Ed Black and the IT and practice technology teams put on their investment banker hats. "We phone them up every few weeks and say, 'Tell us about your updates,'" said Black, the firm's technology strategy leader.
Before a tool can move to testing, it must pass a security and risk audit; only "qualified vendors" make it to the next phase.
From there, testing is twofold. First, a technical evaluation by the firm's technology team aims to ensure the product works as promised. Then a second round with lawyers examines usability and actual value in practice.
Apply: Ropes & Gray rolled out Harvey firmwide in June, after a year of use with a smaller test group, Black said. The firm has also collaborated with Harvey on a "workflow builder" that lets users design and deploy custom agents — software that can carry out tasks on its own.
Hebbia, an AI agent company focused on professional services, has proven particularly useful to lawyers like Melissa Bender, a partner in the asset management group and cohead of the private funds practice.
When institutional investors need fund documents reviewed, Bender uses Hebbia to extract key terms and speed up summaries. She estimates the process now takes two to three hours, less than what would typically be a 10-hour matter.
Align: Black stresses responsible use of the tools, starting with the principle that the results of using these tools are first drafts, not the final product.
The private funds practice requires tool-specific training for junior and mid-level associates, Bender says, while more senior lawyers are "strongly encouraged" to take the training. The goal is to ensure lawyers know how to use the tools appropriately and empower them to speak with clients about the firm's technology capabilities.
"We are in the business of selling legal services," Bender said. "I want our associates to understand the differentiated nature of our offering."
Morgan Lewis requires staff to get credentialed before they can use the tools
Assess: At Morgan Lewis, the first step in adopting AI isn't picking the tool. It's diagnosing the problem, said attorney Timothy Levin, who leads the firm's investment management practice.
Understanding how legal work can be improved with AI is important to ensure tools are applied where they can have a real impact, rather than just throwing tech at a problem, Levin said.
Once a tool passes security and risk checks, it's piloted by an attorney and C-suite advisory group spanning 15 practice areas and firm operations — a cross-section designed to vet the tool's value across the firm's legal work.
Apply: Morgan Lewis has been inundated with startup pitches, says Colleen Nihill, its chief AI and knowledge officer, as the legal tech gold rush draws a wave of new founders. To cut through the noise, Morgan Lewis favors larger enterprise partners that align with its technical standards.
For example, Thomson Reuters is a strategic partner. The firm's advisory group meets regularly with Thomson Reuters to review existing tools, preview the product road map, and beta test unreleased features. They also collaborate to co-develop tools tailored to Morgan Lewis's needs.
One use case at Morgan Lewis involves reviewing fund documents for institutional investors, where CoCounsel Core helps attorneys summarize key terms and flag client-specific dealbreakers.
Align: Nihill said the firm requires its staff to get credentialed for tools before they can use them. Partners and firm leadership were the first to get CoCounsel Core-certified, a process that included Coursera-based coursework, hands-on exercises, and a final assessment.
Once certified, users receive a digital badge displayed on their internal profiles. Nihill says this signals to associates that these tools aren't just approved; they're a professional priority for the firm.