Market
JPMorgan Chase signs a strategic partnership with OpenAI to use Codex and its cyber models, announced by Greg Brockman, with no terms disclosed
OpenAI's Greg Brockman said JPMorgan will use Codex and OpenAI cyber models. No amount, duration or model names were given.
OpenAI president Greg Brockman said on November 4, 2026, at the Wall Street Journal's Tech conference, that JPMorgan Chase has signed a strategic partnership to use Codex, OpenAI's coding agent, and its cybersecurity models. The framing needs a limit: neither company has disclosed the amount, the duration or which cyber models are involved, and Brockman's own post speaks of exploring how AI can support the bank, not of a finished deployment. Disclosure: Anthropic makes Claude, which helps draft articles on this site.
The essentials
- What was announced: a strategic partnership under which JPMorgan will use Codex for engineering and OpenAI's cyber models for security, plus knowledge-work use across the bank.
- Who said it: Greg Brockman, at the WSJ Tech conference, and in an X post that describes the aim as exploring support for engineers, security teams, employees and customers.
- What is missing: the amount, the duration, the names of the cyber models and the scale of the rollout. Some outlets link cyber access to OpenAI's Daybreak program, a point we could not check at the source.
- Context: JPMorgan has used OpenAI models through its internal LLM Suite since 2024 and, per MLQ.ai, has worked with Anthropic on Project Glasswing since April 2026, which continues.
What is new and what is not
The bank already used OpenAI models internally, so the news is a formal partnership extending to coding and security work, not a first contact. Reports describe the cyber side as the part Brockman stressed most, but gave no detail on what the models would do inside the bank.
Reading a partnership announcement
A strategic partnership announced from a conference stage can mean anything from a pilot to a large contract. Without terms, the safest reading is intent. Both firms handle regulated data, so security, audit and data-handling conditions will matter, and none have been described.
What is not established
- Contract value and length: not disclosed.
- Which OpenAI cyber models JPMorgan gets: not named by Brockman.
- How many engineers or systems will use Codex: not stated.
- Whether the Anthropic work on Glasswing changes: no sign of it.
Summary table
| Point | What is stated as of November 7, 2026 | Source |
|---|---|---|
| Announced | November 4, 2026, WSJ Tech conference | Finextra |
| Announced by | Greg Brockman, OpenAI president | PYMNTS |
| Scope | Codex, OpenAI cyber models, enterprise knowledge work | MLQ.ai |
| Terms | Not disclosed by either company | Crypto Briefing |
| Stated aim | Explore how AI can support engineers, security teams, staff and customers | X (Greg Brockman) |
FAQ
How much is the JPMorgan-OpenAI deal worth?
Not disclosed. No amount or duration was given.
Which OpenAI models will JPMorgan use?
Codex and OpenAI cyber models; the specific cyber models were not named.
Does this replace JPMorgan's work with Anthropic?
Nothing reported says so; MLQ.ai notes the Glasswing work with Anthropic continues.
Method and limits
Sources checked on November 7, 2026. We read outlets relaying the WSJ Tech remarks and Brockman's post excerpt, not the full conference session. Disclosure: Anthropic makes Claude, which helps draft articles on this site. This is not investment or security advice.
Sources
- Finextra: JPMorgan Chase signs cybersecurity and coding deal with OpenAI (verified 2026-11-07)
- PYMNTS: OpenAI forms JPMorgan partnership ahead of 2027 IPO (verified 2026-11-07)
- Crypto Briefing: OpenAI and JPMorgan AI cybersecurity partnership (verified 2026-11-07)
- MLQ.ai: JPMorgan Chase signs OpenAI deal for Codex and cybersecurity models (verified 2026-11-07)
- FourWeekMBA: Brockman reveals JPMorgan deal puts enterprise above 50% (verified 2026-11-07)
- X (Greg Brockman): Post by Greg Brockman on the JPMorgan partnership (verified 2026-11-07)
Spotted an error in this article? Report it · Our corrections policy
