Goldman Sachs has reportedly pulled access to Claude Code for its bankers and traders, according to The State of AI. The move marks a sharp reversal for Anthropic's enterprise ambitions, just months after the tool appeared to be gaining ground inside one of Wall Street's most influential institutions.
The reported decision comes after the investment bank had already begun rolling out Claude to employees earlier this year. Removing Claude Code from traders and bankers signals rising caution around how generative AI tools handle sensitive financial data, and it raises fresh questions about how quickly major banks will hand coding assistants to non-technical staff.
Goldman Sachs pulls Claude Code: what we know
According to The State of AI, Goldman Sachs has cut Claude Code access for the bankers and traders who had been using it. The report does not specify the exact date of the decision, the number of staff affected, or the full reasoning behind the pullback. Anthropic and Goldman Sachs have not issued detailed public statements clarifying the situation.
The move lands in an awkward moment for Anthropic. Claude Code was positioned as a flagship enterprise product, designed to let users write, debug and automate code through a conversational interface. Losing traction inside Goldman Sachs, even partially, undercuts the narrative that Claude is the default AI assistant for high-stakes financial work.
Why banks are wary of AI coding tools
Financial institutions operate under strict compliance, data-residency and audit rules. Any tool that can read internal documents, execute commands or touch production systems raises red flags for risk teams. Even when a vendor promises enterprise-grade controls, banks often prefer to limit the blast radius by restricting who can use the tool and what it can access.
Goldman's decision may also reflect a broader pattern seen across Wall Street, where AI pilots are increasingly moving from open experimentation to tighter, role-based access. Coding assistants aimed at developers are one thing. Handing similar tools to traders and bankers is a different risk calculation entirely.
ماذا يعني هذا لك؟
For Tunisian readers, the Goldman Sachs story is a reminder that enterprise AI adoption is not a straight line. Tools that look unstoppable in one quarter can be pulled back the next, especially when regulators, auditors and internal risk teams get involved. Companies evaluating Claude Code or similar assistants should expect stricter governance, not looser.
