AT&T uses OpenAI to help lawyers access decades of legal knowledge faster
A new system would let AT&T's lawyers get answers faster by pulling from past cases and research — but humans still make every decision.

How it works
An AI assistant that helps lawyers quickly find past cases and research by typing questions in plain English.
- Type your question in normal language
- System finds related past cases and research
- Lawyer gets recommended next steps
- Lawyer makes the final decision
What you could do with it
- A lawyer at AT&T could type a question in everyday language, like asking about reviewing a contract for a new vendor partnership, and the system would automatically find past similar cases and relevant research. 1
- AT&T's legal intake system could classify and prioritize incoming requests from across the business, connect them with relevant precedent, and give the responsible lawyer historical context and recommended first steps. 1
- AT&T's legal department could expand its ability to provide legal judgment at greater scale, letting lawyers handle more work without sacrificing quality. 1
A smarter "front door" for legal requests
AT&T's legal department employs hundreds of lawyers and legal professionals around the world. When a request comes in from another part of the business, the legal team needs to quickly understand the issue, gather relevant context, and decide what matters most.
Bill Ryan, AT&T's Chief Compliance Officer, described one upcoming project as an intake system with "a front door where we have natural language intake." 1
In plain terms, this means a lawyer could type a question in everyday language — like "We need to review a contract for a new vendor partnership in Europe" — and the system would automatically find past similar cases, relevant research, and recommend next steps. 1
Building on previous AI work
The partnership grew out of LegalEdge, a pre-existing AT&T team that united attorneys, legal workers, technologists, and project managers who work on using AI for significant legal cases. 1
David McAtee, AT&T's Senior Executive Vice President and General Counsel, said the collaboration with OpenAI aims to "build that plumbing that powers the knowledge layer throughout the company and that all of us will draw into." 1
The goal is to connect more of the company's accumulated legal expertise — decisions made years ago, contracts drafted earlier, lessons learned from past disputes — directly to the work landing on a lawyer's desk today. 1
Guess
Who makes the final legal decisions?
The decision making always remains with the lawyers, not the AI.
Humans stay in control
Both leaders stressed that AI would assist but not replace legal judgment.
Ryan said what excites him most is "expanding our ability to provide legal judgment at greater scale" — meaning lawyers could handle more work without sacrificing quality. 1
McAtee put it plainly: "The decision making will always remain with the lawyers." 1
The AI would handle the research and organization, giving lawyers better information so they can spend more time applying their expertise.
What this means for the wider world
AT&T is one of more than one million businesses worldwide that OpenAI says are achieving meaningful results with its tools. 1
The company's legal department case shows a growing trend: big organizations are moving beyond trial runs with AI and building systems that connect historical knowledge to daily work — particularly in fields that require deep expertise and careful judgment.
This story reports on a trend for information only. It is not advice or encouragement to do any of it.
Key facts
- AT&T is partnering with OpenAI to enhance its in-house legal department
- A planned intake system will let lawyers describe issues in plain language
- The project builds on LegalEdge, AT&T's existing AI-for-legal team
- Human lawyers will make all final decisions; AI handles research and organization
Sources
Keep reading
New modelsOpenAI launches Decisions API for fast yes no AI evaluationsHerald of AINew modelsOpenAI launches Decisions API for fast yes no AI evaluations
A new API turns complex text and image analysis into simple answers — and runs about ten times faster than the Responses API.





