Goldman Sachs Limits AI Coding Tools to Engineers

Most non-engineering staff lose access to Anthropic's Claude and Microsoft GitHub under updated internal guidelines announced in October.

Updated on Nov. 2, 2026 in Artificial Intelligence

Goldman Sachs Limits AI Coding Tools to Engineers

Live Poll

Should workers in your job get the same AI tools as specialists?

Goldman Sachs announced updated internal guidelines in October that restrict access to advanced AI coding tools, mainly to engineering staff. The affected tools include Anthropic's Claude and Microsoft GitHub, which non-technical employees had previously been able to use.

Why it matters

The restrictions follow a trial period in which the bank weighed efficiency gains against risks and controls, reflecting how major financial institutions are balancing AI adoption with oversight. Access is now limited by job type rather than location.

Goldman Sachs is rolling out its proprietary GS AI Assistant to approximately 46,000 employees while running ongoing trials of agentic AI, including in trade surveillance. Traders had used the restricted tools for trade monitoring, model building and content generation, and the bank has not disclosed the exact number of staff affected.

The players

Goldman Sachs

Goldman Sachs is a global investment bank and financial services firm headquartered in New York that has been expanding its internal use of artificial intelligence.

Anthropic

Anthropic is an artificial intelligence company that develops the Claude family of AI models and was founded by former OpenAI researchers.

Microsoft

Microsoft is a multinational technology company that owns GitHub, a widely used platform for software development and code collaboration.

The details

Access to Anthropic's Claude and Microsoft GitHub is now restricted by job type rather than location, with engineering staff retaining access. Quant strategists are exempt. A Goldman spokesperson said the adjustments followed a trial period, and the bank had earlier restricted Claude for Hong Kong-based bankers in April after a contract review.

April 2026: Claude access restricted for Hong Kong-based bankers following a contract review.
October 2026: Updated internal guidelines restricting advanced AI coding tools announced.

The Tech Race

Goldman's move to restrict Anthropic's Claude follows its April decision to limit the same tool for Hong Kong-based bankers, extending a pattern of tightening access as banks test generative AI. The change places Goldman's in-house GS AI Assistant, now rolling out to about 46,000 employees, against third-party tools from Anthropic and Microsoft.

Employees outside engineering who relied on Claude or GitHub for daily work will need to request alternative tools or use the GS AI Assistant instead. The change signals that access to AI productivity tools at large banks may increasingly depend on job function rather than individual preference.

The takeaway

Large financial firms are moving toward role-based AI access, pairing in-house tools with tightly controlled third-party models. Staff in non-technical roles should expect narrower access to outside AI tools as institutions balance productivity against risk.