Full Breakdown
Goldman Sachs Bars Hong Kong Bankers from Anthropic’s Claude AI Model
4/29/2026, 11:19:00 AM
Core Event
On 28 April 2026 Reuters reported that Goldman Sachs removed access to Anthropic’s Claude model for its Hong Kong bankers. The bank cited a strict reading of its contract with Anthropic. Other AI tools such as Gemini and ChatGPT remain on the internal platform. Goldman declined comment.
Background & Context
U.S. AI models are barred from mainland China, while Hong Kong remains open. OpenAI limited API access for China in 2024, and the HKMA warned banks to assess cybersecurity risks from advanced AI. Goldman had embedded Anthropic engineers to build agents for accounting and onboarding.
Key Figures & Groups
Key actors include Goldman Sachs, the U.S. investment bank that barred Claude; Anthropic, the model’s developer; Marco Argenti, Goldman’s chief information officer; the Hong Kong Monetary Authority (HKMA); and the Financial Times, which reported the contract-interpretation quote.
Timeline
Why It Matters / Impact
The cut-off may weaken Hong Kong’s finance-centre status if bankers lose a tool for coding, modeling and summarising documents. Industry analysts warn a gap could emerge versus peers with AI access. The case signals a shift toward AI with controls, raising costs while aiming to curb model-risk and data-leakage.
Official Statements & Responses
Anthropic said Claude has never been supported in Hong Kong. Goldman declined comment, saying restriction follows interpretation. HKMA and Hong Kong government did not respond. Argenti noted collaboration with Anthropic, indicating narrows but does not end partnership.
Criticism & Opposition
Industry observers warn the restriction could erode Hong Kong’s finance-centre status if rivals keep AI tools. Reliance on third-party AI developers poses structural risk, as provider changes or outages can disrupt workflows. Some analysts see the move as pre-emptive to avoid regulator-mandated limits.
Conflicting Reports & Gaps
Reuters could not confirm whether other banks have limited Anthropic access or if other AI vendors face similar bans. Anthropic’s public API list does not include Hong Kong, yet the precise contractual language remains undisclosed. The HKMA’s future stance on generative-AI regulation has not been publicly outlined.
Verbatim Quotes
- “Goldman Sachs strictly interpreted the terms of its contract with Anthropic” — Financial Times source
- “concluded that personnel in Hong Kong should not use Anthropic's products.” — Financial Times source
- “Claude has never been officially supported in Hong Kong,” — Anthropic spokesperson
- “Goldman Sachs has quietly restricted employee access to Anthropic’s Claude in its Hong Kong offices, a move that signals a cooling trend in how the world’s most sophisticated financial institutions are managing the reality of generative AI adoption.” — Startup Fortune
- “The restriction also highlights a deeper, more structural risk: the reliance on third-party AI developers.” — Startup Fortune
What’s Next
The HKMA is expected to issue clearer AI-model risk guidance in the coming months. Goldman may broaden its tiered-access framework, limiting generative AI to vetted, secure environments while continuing to evaluate Anthropic’s offerings. Other banks are likely to review their AI-governance policies, potentially prompting sector-wide restrictions.
