Submit

Goldman Sachs pulls Claude Code from traders and bankers

Goldman Sachs limited Claude Code and GitHub Copilot to its roughly 12,000 engineers in October 2026, cutting off traders and bankers, Bloomberg reports.

Written by AiAgentsListing Team

Goldman Sachs pulls Claude Code from traders and bankers

What happened

Goldman Sachs has limited Anthropic's Claude Code and Microsoft's GitHub Copilot to its engineers, cutting off the traders and bankers who had been using the coding agents on the job. Bloomberg News reported on 2 November 2026 that internal guidance updated in October restricts the tools to Goldman's roughly 12,000 engineers, out of about 47,000 total staff. The bank's own chief information officer had praised non-engineers' "vibe coding" with the same tools as recently as July.

Goldman Sachs limited Anthropic's Claude Code and Microsoft's GitHub Copilot to its roughly 12,000 engineers in October 2026, cutting off the bank's traders and bankers who had used the coding agents to track client flows and build market models. The bank cited productivity and risk concerns, Bloomberg News reported on 2 November 2026.

What changed

  • Internal guidance updated in October 2026 restricts Claude Code and GitHub Copilot to Goldman's engineers, about 12,000 of its roughly 47,000 staff, according to Bloomberg's 2 November 2026 report.
  • Traders and bankers who used the tools to monitor client trading flows, model market moves and positions, and put together written and graphic material lost access.
  • Strats, who work across trading, quantitative research and software engineering, still have access, a person familiar with the matter told Bloomberg.
  • Goldman CIO Marco Argenti had talked up trader-built "vibe coding" as recently as July 2026, saying non-technical employees built tools with AI and passed them to engineers to improve.
  • Goldman spokesperson Nick Carcaterra said the bank makes certain AI tools available to staff "to determine the best use cases" across the firm.
  • The change follows a narrower, earlier restriction: Goldman barred its Hong Kong bankers from Claude in April 2026 over contract terms, and JPMorgan removed Claude from its own Hong Kong approved list in June.

Why it matters for agent builders

Code an engineer writes with a coding agent usually goes through review, testing and deployment controls before it touches anything important. A script a trader builds to watch client flows can end up in daily use without going through any of that, and banks are required to keep information barriers between groups, for example between bankers working on a deal and traders who could profit from knowing about it. A tool assembled by someone outside engineering was not built with those controls in mind, and an engineer can only check it if someone sends it their way.

Bloomberg's sources said productivity was also part of Goldman's review. Agent usage is billed by the token, and the restriction arrives while Goldman and Morgan Stanley are lead underwriters on Anthropic's own IPO: Anthropic filed its S-1 on 26 October 2026 and is expected to list on Nasdaq under the ticker ANT before the end of November. For anyone shipping a coding agent into a regulated workplace, the lesson is to settle who owns agent-written code, and route it through the review and entitlement rules that already apply to human-written code, before handing the agent out widely.

What to watch

Goldman and Barclays are now pulling in opposite directions. Barclays said in October it expects half its developers to be using Claude Code by the end of the year, and three weeks later named developers among 5,000 job cuts; unlike Goldman, it is using coding agents to shrink its engineering headcount rather than to replace non-engineers' spreadsheets. Whether other banks copy Goldman's role-based restriction, and whether it survives once Anthropic's IPO roadshow is behind it, is the open question.

Key takeaways

  • Goldman Sachs restricted Claude Code and GitHub Copilot to its roughly 12,000 engineers in October 2026, per internal guidance Bloomberg reported on 2 November.
  • Traders and bankers lost access to tools they had used to monitor client flows and model market positions; strats kept it.
  • Goldman's CIO had praised the same trader-built "vibe coding" tools in July 2026, months before the restriction.
  • Goldman and Morgan Stanley are lead underwriters on Anthropic's IPO, which is expected to list on Nasdaq under ticker ANT before the end of November 2026.
  • Barclays is moving the opposite way, expanding Claude Code to developers while cutting 5,000 jobs including developer roles.

FAQ

Which AI coding tools did Goldman Sachs restrict?

Goldman restricted Anthropic's Claude Code and Microsoft's GitHub Copilot. Internal guidance updated in October 2026 limits both tools to the bank's engineers; Goldman's separate, in-house AI assistant, which draws on models from several companies, remains available firmwide.

Who at Goldman Sachs can still use Claude Code?

Goldman's roughly 12,000 engineers keep access, as do strats, who work across trading, quantitative research and software engineering. Traders and bankers outside those groups lost access under the October 2026 guidance.

Why did Goldman Sachs restrict AI coding agents for traders and bankers?

Bloomberg's sources pointed to a review of productivity and risk concerns. Code written by a trader or banker was not subject to the review, testing and information-barrier controls that apply to engineer-written code, and Bloomberg's sources said cost was also part of the review.

Source: State of AI Discussion: Hacker News

Developers comparing coding agents before adopting one at scale can browse the coding category on AI Agents Listing.

Share:

Subscribe to our newsletter

One email a week. New agents, MCP servers and skills, and what is actually getting traction.

More news

  1. Anthropic details how Claude's agent memory system works

    Anthropic's 6 November 2026 research post explains how Claude agents build, prune and correct memory across sessions through a process it calls dreaming.

  2. Perplexity ships Teammates, agents for ongoing work

    Perplexity launched Teammates on 6 November 2026, agents built on its Computer platform that own ongoing work like sales follow-ups, not single tasks.

  3. Runlayer ships MCP Explained, an interactive protocol guide

    Runlayer's free interactive guide, MCP Explained, walks through every JSON-RPC message in the Model Context Protocol, including the 2026-07-28 spec.