US banks roll out American AI in Hong Kong despite geopolitical tensions

US banks push American AI into Hong Kong

Leading American banks are pushing advanced AI tools built on US technology into their Hong Kong operations, even as Beijing‑Washington tensions complicate access to some top-tier software. The move underscores how global finance is racing to deploy generative AI for sales, risk and operational efficiency — and asks a blunt question: how far will banks go when access to the best models is uneven across markets?

Citigroup pilots an in‑house agent called Arc

It has been reported that Citigroup Hong Kong this month began testing an in‑house AI agent platform called Arc, initially aimed at software developers and engineers. Senior management has reportedly been “aggressively” encouraging AI adoption since last summer; an anonymous employee told reporters that Arc enables more end‑to‑end automation, turning hours of manual data gathering and scenario modelling into a single agent‑driven workflow.

Restricted US models and a compliance minefield

It has been reported that US AI firms OpenAI and Anthropic have not made their models available to users in Hong Kong and mainland China, forcing banks to navigate a compliance minefield. That reality is driving banks toward internal platforms, localised deployments and legal workarounds to keep using advanced tooling without breaching export controls, data‑residency rules or sanctions.

Why this matters

The trend matters beyond product rollouts. Trade policy, export controls and geopolitical rivalry are increasingly shaping which AI systems global banks can use and where. If access to leading US models remains constrained, international lenders may invest heavily in bespoke, regionally hosted AI or lean on non‑US providers — with implications for security, regulatory oversight and Hong Kong’s role as Asia’s international finance hub.