Goldman Sachs is limiting access to Anthropic's Claude Code and Microsoft's GitHub coding tools to its engineers, ending a stretch of several months in which bankers and traders at the firm could use them for their own work. The change follows updated internal guidance Goldman issued in October, which came to light on Monday.
Under the new rules, access is set by job type rather than by location. Goldman's roughly 12,000 engineers keep the tools, and the firm's quantitative strategists, known inside the bank as strats, are exempt from the restriction.
Goldman has not disclosed how many employees lose access. Traders had used Claude Code and GitHub to monitor client trading flows, build models, and produce written and graphic materials, and some of them are frustrated by the change.
Key facts
- 01Goldman Sachs is limiting Anthropic's Claude Code and Microsoft's GitHub tools to engineers under October internal guidance.
- 02About 12,000 Goldman engineers keep access, and the firm's quant strategists are exempt.
- 03Traders had used the tools to monitor client trading flows and produce written and graphic materials.
- 04Goldman is rolling out its in-house GS AI Assistant to about 46,000 employees.
- 05In July, CIO Marco Argenti praised "vibe coding" by non-technical staff.
The restriction marks a turn from the tone Goldman set earlier this year. In July, Chief Information Officer Marco Argenti praised "vibe coding" by non-technical staff, whose prototypes were passed to the firm's engineers to refine. The new guidance followed a trial period in which the bank weighed efficiency gains from the tools against risks and controls.
Goldman described the tools as part of an ongoing effort to find where AI fits best inside the firm.
“We make certain AI tools available to our people to determine the best use cases across different parts of the firm.”— Nick Carcaterra, Goldman Sachs spokesperson
Goldman is not pulling back from AI more broadly. The bank is rolling out its in-house GS AI Assistant to about 46,000 employees, and staff outside engineering who relied on Claude Code or GitHub are being directed to request alternatives or use that assistant.
The firm is also running agentic AI trials, including in trade surveillance, so the shift is about which tools sit in which hands rather than whether Goldman uses AI at all.
Other large banks are drawing their own lines. JPMorgan Chase CEO Jamie Dimon has warned about "rogue agents" and said outside AI agents would not be allowed inside the bank. Bank of America has limited agents' access to consumer deposits while it develops liability rules for AI-authorized payments, and Citigroup does not let third-party agents log into accounts or make transactions, according to reports.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
