AT&T outlines AI system to prioritize legal requests in OpenAI collaboration

AT&T wants AI to help determine which legal requests need attention and assemble the company knowledge lawyers need to address them. In an October 21 account published by OpenAI, executives described a planned natural-language intake system, adding detail to a collaboration announced earlier this month. Legal decisions would remain with lawyers.

From LegalEdge to a department-wide effort

The starting problem is the work that happens before legal advice. AT&T’s legal department includes hundreds of lawyers and legal professionals worldwide. They must gather facts, documents and relevant past work, sometimes relying on expertise spread across different teams, before they can assess a business request.

AT&T established LegalEdge in January 2026 as an AI-first center of expertise. The company says it handled nearly 100 matters in its first six months. Its team brings together attorneys, other legal professionals, technologists and project managers, with a focus on applying AI to major legal matters.

In its October 8 announcement, AT&T named LegalEdge as the foundation for its OpenAI collaboration. The companies said they would design, build and evaluate practical legal workflows under attorney oversight. Their stated goals included making institutional knowledge easier to access, supporting appropriate routine work and freeing legal professionals for strategy, advocacy and other work requiring judgment.

OpenAI’s legal industry general manager, Jason Boehmig, framed the collaboration around the legal team’s role in business decisions. He identified evaluating new opportunities and helping teams make informed decisions and act on them as areas where AI could help in-house lawyers contribute more. Those are ambitions for the work, not reported outcomes.

A request arrives before a lawyer starts

The October 21 account gives that effort a more concrete shape. Chief Compliance Officer Bill Ryan described an intake system that would accept requests in natural language, then classify and prioritize them. Intake is the stage at which the team identifies the issue, gathers context and decides what needs attention first.

The planned system would connect a request with relevant precedent, research and previous work. It would then give the responsible legal professional historical context and recommended first steps. The intended benefit is a better-informed starting point, with less time spent assembling the material needed to assess a matter.

General Counsel David McAtee described the larger aim as building a knowledge layer that people across the company could draw on. That connects the intake proposal to the partnership’s original purpose: making AT&T’s accumulated legal knowledge and expertise more accessible to the business, rather than improving only an individual lawyer’s productivity.

The decision making will always remain with the lawyers.

David McAtee, AT&T senior executive vice president and general counsel

Lawyers set the agents’ role

AT&T’s original announcement also described different levels of AI involvement, with lawyers determining how agents are used. It identified two possible roles:

• For some tasks, such as frequently asked questions, an agent may provide a response.
• For others, agents may gather and combine relevant information to help lawyers provide legal advice.

The company says governance, confidentiality, accountability and human review will be built into every workflow. Those are stated design commitments for the collaboration. The October 21 account still describes the intake system as planned; it does not announce a released product or provide measured productivity results.

Evaluation is part of the announced work, and Ryan has set out how AT&T intends to judge it. Success, he said in the October 8 announcement, should mean better quality, more accessible knowledge, reusable capabilities and more capacity for work requiring experience and judgment—not simply deploying more technology.