Goldman Sachs has recently restricted the use of Claude AI in Hong Kong, thus opening a new chapter in the relationship between banks, technology, and geopolitics. Artificial intelligence therefore becomes a strategic issue for global finance as well.
According to reports, access to the system developed by Anthropic was initially available through the bank’s internal infrastructure.
However, in recent weeks, Goldman Sachs has decided to discontinue its use for staff based in Hong Kong, while keeping other tools such as ChatGPT and Gemini active.
It therefore appears that this may be a move not against artificial intelligence in general, but only a selection of which platforms to use.
Summary
Why did Goldman Sachs block Claude AI?
The decision seems to be based on a contractual assessment, namely that Anthropic does not officially include Hong Kong among the areas supported for the use of its services.
A legal gray area that a global bank like Goldman Sachs can hardly afford to ignore.
In a highly regulated financial environment, even seemingly technical details can become decisive. The use of a service not formally authorized in a given jurisdiction can expose institutions to compliance, audit, and legal liability risks.
And when it comes to institutions like Goldman Sachs, risk management concerns not only cybersecurity, but also consistency with international and local regulations.
The case becomes even more interesting when considering Hong Kong’s position. The city represents a balancing point between the global financial system and the growing influence of mainland China.
Unlike China, where many Western AI tools are restricted or banned, Hong Kong has in fact maintained a certain degree of openness. However, this openness is not absolute and often depends on the choices of individual tech companies.
The fact that Anthropic does not include Hong Kong among the officially supported regions creates a particular situation. This means that local companies can access global technologies, but only within boundaries defined by the providers themselves.
And the decision by Goldman Sachs, as mentioned,fits exactly into this gray area.
AI under control: the role of regulators
Another key element concerns the growing interest of financial regulators in AI technologies.
The Hong Kong Monetary Authority has already begun contacting banking institutions to gather information on the use of new models, including Anthropic’s Mythos.
This is a decisive shift, as it means that artificial intelligence is no longer seen only as a tool for efficiency, but as a potential source of systemic risk.
It is no coincidence that banks already use AI for analysis, automation, and decision support. However, increasingly advanced models introduce new vulnerabilities, both in terms of cybersecurity and data management.
For this reason, authorities are asking institutions to continuously update their risk frameworks.
Taking a step back, note how the fact that ChatGPT and Gemini remain accessible while Claude is blocked highlights a fundamental point. Banks are not rejecting AI, as already mentioned, but are differentiating between providers and solutions.
As we know, each platform has different characteristics in terms of security, geographic support, transparency, and integration. Decisions are therefore not only technological, but also strategic.
In the case of the Claude AI block, the issue is therefore not only the quality of the model, but the context in which it is used.
This opens a new phase in the artificial intelligence market: competition will not only be between more advanced models, but also between platforms that are more compatible with global regulatory requirements.
The weight of geopolitics
Behind this choice one can also glimpse the growing tension between the United States and China in the field of AI. In recent months, control over advanced technologies has in fact become a central issue in international relations.
Washington has restricted China’s access to advanced chips and critical infrastructure. Beijing, in turn, has tightened control over startups and technology transfers. In this context, artificial intelligence becomes a strategic tool, not just an economic one.
Hong Kong, finding itself in the middle of these dynamics, becomes a potential balancing point. And companies must move cautiously, avoiding exposure to political or legal risks.
In any case, Goldman Sachs’ decision regarding Claude AI could have ripple effects. Other financial institutions could indeed adopt similar policies, assessing more carefully which AI tools to use and where.
This could slow the adoption of artificial intelligence in the banking sector, pushing toward more controlled and internally integrated models.
At the same time, it could foster the development of AI solutions designed specifically for regulated environments, with greater focus on compliance and security.
The case also highlights a fundamental tension. AI promises efficiency, automation, and competitive advantages, but its use also entails new risks that are difficult to predict and manage.
Banks, by nature, tend to favor stability over speed. This leads them to move cautiously, even at the cost of temporarily giving up advanced technologies.
In the long term, however, this prudence could turn into an advantage. In a context where AI is increasingly central, having reliable and compliant systems could make the difference.
