JPMorgan, Goldman Restrict Claude Access in Hong Kong 

• JPMorgan and Goldman restrict Claude in Hong Kong amid AI security concerns.
• Regulators and banks tighten AI controls as cyber threats and policy risks grow.
• Global banks balance AI adoption, compliance demands, and growth strategies.

Major financial institutions are tightening their approach to artificial intelligence as regulatory uncertainty and national security concerns reshape how banks deploy advanced AI tools. JPMorgan Chase recently blocked employees in Hong Kong from accessing Anthropic’s Claude models, joining Goldman Sachs in limiting the technology’s availability in the region. The move highlights growing caution among global banks as governments, regulators, and technology companies reassess the risks associated with increasingly powerful AI systems.

Banks Tighten Controls on AI Platforms

JPMorgan removed Anthropic’s models from the list of approved large language models available to its Hong Kong workforce. The decision followed concerns surrounding Anthropic’s licensing terms, which exclude usage across Greater China, including Hong Kong.

Goldman Sachs adopted a similar position earlier this year. Consequently, employees in Hong Kong no longer have access to Claude through the bank’s approved technology platforms.

The restrictions arrive amid broader concerns about advanced AI capabilities. Recently, US authorities pushed Anthropic to suspend access to one of its most sophisticated models after officials raised security concerns. Policymakers reportedly worried that highly capable AI systems could expose software vulnerabilities or accelerate cyber threats.

Additionally, technology companies continue to express concerns about model distillation. This practice allows developers to train competing systems by extensively studying the outputs of leading foreign AI models. As a result, several AI providers have increased geographic restrictions and tightened access controls.

Financial Sector Warns of Emerging Risks

While banks evaluate AI-related threats, Wall Street institutions also face regulatory challenges. Industry groups recently warned that proposed Basel Endgame capital requirements could increase capital charges for certain trading activities by between 30% and 89%.

Financial organizations argue that higher capital requirements may reduce liquidity in the $29 trillion US Treasury market. Moreover, the concerns come as regulators prepare for expanded Treasury clearing requirements later this year.

Elsewhere, Japanese banking leaders have highlighted the growing cybersecurity risks linked to advanced AI technologies. Executives warned that sophisticated attacks could force temporary service interruptions, including online banking platforms and automated teller machines.

Global Banking Landscape Continues to Shift

Regulatory developments also emerged outside the AI sector. In Australia, HSBC agreed to a proposed A$35 million penalty after regulators identified weaknesses in its anti-scam controls. Authorities found that customers faced elevated risks from unauthorized transactions during a one-year period.

Meanwhile, a US federal judge approved the dismissal of a long-running sanctions case against Turkey’s Halkbank. The lender had faced allegations involving approximately $20 billion in restricted Iranian funds.

Across Africa, Standard Bank continues pursuing growth opportunities. The lender estimates that expanding intra-African trade could unlock a revenue opportunity worth roughly R250 billion, or $15.4 billion. Therefore, the bank plans to strengthen its presence in key markets, including Nigeria, Ghana, Kenya, Uganda, and Tanzania.

Related: Anthropic Disables Top AI Models After U.S. Order Blocks Foreign Access