JPMorgan Chase has quietly pulled Anthropic's Claude from the menu of artificial-intelligence tools its Hong Kong employees are allowed to use — a decision driven not by the model's performance but by the fine print of a licensing contract, and one that makes the largest U.S. bank the second Wall Street giant in roughly two months to wall off the same vendor in the same city.
The bank removed Claude from the internal drop-down list of approved large language models available to staff in the territory, according to a Financial Times report cited by multiple outlets on June 18. The trigger, per the reporting, was the wording of Anthropic's usage terms in its licensing agreement with JPMorgan — language that created enough jurisdictional ambiguity for the bank's compliance teams to remove the tool rather than risk getting it wrong. Employees elsewhere in JPMorgan's global network can still use Claude; only Hong Kong has been carved out.
What happened
The move closely mirrors Goldman Sachs, which removed Claude from the approved-tools list for its Hong Kong-based bankers in late April 2026. Goldman's staffers had been able to reach Claude through an internal AI platform until the bank took what was described as a strict interpretation of its Anthropic contract — concluding, after consulting the company, that its Hong Kong employees should not be able to use any Anthropic products. Rival models including Google's Gemini and OpenAI's ChatGPT reportedly remained available on Goldman's platform, underscoring that the restriction was vendor-specific rather than a blanket retreat from AI.
The common thread is Anthropic's own terms of service. The company's licensing language excludes usage across Greater China, including Hong Kong, on the grounds of what it has publicly called "legal, regulatory, and security risks." In September 2025, Anthropic went further, updating its terms to bar any entity more than 50% owned by Chinese companies from using Claude — regardless of where that entity is registered — explicitly to stop subsidiaries incorporated in Hong Kong or Singapore from routing around the rules. In Anthropic's framing, companies "subject to control" from authoritarian jurisdictions can be compelled to share data or cooperate with intelligence services, "regardless of where they operate."
For a global bank running a major desk in a Chinese territory that is also one of the world's premier financial centers, that language turns a productivity tool into a compliance question. Removing Claude in Hong Kong is the cautious answer — the kind institutions reach for when the cost of misreading a contract is regulatory rather than commercial.
The backdrop: export controls and a hardening U.S. line
The bank-level decisions sit inside a much larger shift in Washington's posture toward advanced AI and China. U.S. Commerce Secretary Howard Lutnick ordered Anthropic to suspend exports of its most capable models — the Mythos and Fable systems — including to foreign nationals, citing the risk that frontier American models could be diverted to military or intelligence end users in China, Russia, and other "countries of concern." That intervention reframed the most powerful U.S. models as something closer to controlled dual-use technology than ordinary software.
Hong Kong sits awkwardly in that picture: a Chinese territory and a global capital hub at once. Any AI tool whose licensing creates doubt about cross-border data flows or who can access it becomes a liability for a bank operating there. Two of the largest U.S. banks independently reaching the same conclusion about the same vendor in the same city is the kind of pattern that tends to precede an industry norm.
Why this matters: the IPO irony
The timing is awkward for Anthropic in a way that goes beyond a single city. The company has lined up Morgan Stanley and Goldman Sachs to lead its initial public offering, with JPMorgan also reported to be working on the deal — meaning the very banks restricting Claude in Hong Kong are the ones preparing to sell Anthropic to public markets. Anthropic filed a draft registration statement with the SEC on June 1, with reports pointing to an October 2026 listing, on the heels of a $65 billion Series H round in May that lifted its post-money valuation to roughly $965 billion.
That valuation rests heavily on the breadth of Anthropic's enterprise reach. The company has spent the past year embedding Claude across finance — from a reported $1.5 billion Wall Street joint venture to pilots inside banks for accounting and compliance work. The Hong Kong restriction does not unwind any of that; it carves a single jurisdiction out of an expanding relationship. But it marks the first highly visible place where geopolitics — not capability or price — sets the boundary of where Claude can be used. Each bank that pulls Claude in a sensitive market is a small dent in the growth story Anthropic is taking to investors, and a reminder that a frontier vendor's addressable market is now shaped as much by export policy as by product quality.
There is a structural lesson here too. As one Upstage executive argued in the wake of Anthropic's curbs, the restrictions effectively make the case for "sovereign AI" — locally controlled models built to sidestep exactly this kind of cross-border licensing risk. For multinational banks, the episode is a preview of an AI strategy that may have to be drawn jurisdiction by jurisdiction rather than rolled out globally.
What to watch next
Three questions will shape how big this becomes. First, whether other global banks with large Hong Kong operations — Morgan Stanley among them — follow Goldman and JPMorgan, which would harden a two-firm pattern into a sector-wide standard. Second, whether Anthropic revises its licensing language to resolve the ambiguity, or treats Greater China exclusion as a permanent feature of its compliance posture ahead of going public. And third, how all of this is disclosed in Anthropic's IPO paperwork: investors reading the S-1 will want to know how much of the company's enterprise footprint is exposed to export controls and region-based restrictions that can shrink its market with a single contract clause.
Neither JPMorgan nor Anthropic has commented publicly on the specifics beyond the FT's account, and it is not yet clear whether the Hong Kong restriction will spread to other jurisdictions. What is established is the fact of it — a second Wall Street bank has pulled Claude from its Hong Kong toolset — and the reason points less at the technology than at the widening geopolitical fault line running beneath it.
"...legal, regulatory, and security risks."-- Anthropic, Company terms, on excluding Greater China usage