OpenAI and JPMorgan Chase ink AI deal spanning code and cyber defense

OpenAI has a new enterprise customer, and it happens to be the biggest bank in the US. Greg Brockman, OpenAI's president, unveiled the partnership at the Wall Street Journal Tech Conference on November 4, 2026, confirming JPMorgan Chase will deploy the company's AI across engineering, cybersecurity and wider enterprise work.

Two products sit at the center of the arrangement. The first is Codex, OpenAI's coding agent. The second is a set of specialized cybersecurity models that JPMorgan plans to run internally to shore up its defenses. The bank's stated ambition stretches further: an enterprise-wide transformation, not just a security patch.

Brockman framed the whole thing around trust. He said hitting regulatory standards is central to this kind of deployment, and he pointed to JPMorgan as a key institution on that front. That framing matters, because watchdogs will want to see how the bank validates AI-generated code and how those cyber models get monitored.

This isn't a first date. JPMorgan has been using OpenAI's models since 2024 through LLM Suite, its in-house platform, which also gives staff access to Anthropic's models. The suite has reached hundreds of thousands of employees and, by early 2026, supported roughly 150,000 weekly active users. The bank says the payoff has been real: its AI initiatives generated about $2 billion in annual savings by early 2026, plus what it describes as substantial productivity gains.

Anthropic isn't getting shown the door. Since April 2026, JPMorgan has partnered with Anthropic on Project Glasswing, a push into defensive cybersecurity with AI. The new OpenAI deal doesn't replace that relationship — but it does put the two labs bidding for the same security budget inside the same bank.

There's a shift inside OpenAI here too. Enterprise revenue has climbed from roughly 30% of the business to a majority share, making deals like this one the main event rather than a side hustle. The $2 billion in annual savings is the number competitors will notice first, and it's the number regulators will likely poke at as these tools move deeper into critical systems.