← All articles · Partners
PLENDE

Goldman Sachs Pulls AI Coding Tools From Traders and Bankers

03.11.2026
This content was prepared with the help of AI.

What Goldman Sachs Changed

Goldman Sachs is pulling AI coding tools such as Claude Code and GitHub Copilot away from traders and bankers. Access is now limited to engineers only. According to The Decoder, the decision comes even though CIO Marco Argenti publicly praised \"vibe coding\" by non-technical staff as recently as July. The bank cites \"productivity and risk management\" as the reason for the reversal.

Why Non-Technical Staff Create New Risk

The problem is not only the tool but who uses it. Code generated by AI and run by someone without engineering training may contain errors that the author cannot detect. In a financial institution, such code can touch reporting, valuation models or compliance processes. The source does not describe any specific incidents, so the exact cause of the decision cannot be identified beyond the bank's general explanation.

How to Roll Out AI Tools Without Losing Control

Firms that want to avoid a similar change of course can work through a few steps:

  1. Map which roles actually need code generation and which use it only occasionally.
  2. Define which data and systems the tool can reach before it is made available.
  3. Require review of generated code by a competent person before anything goes to production.
  4. Monitor usage and keep a change log so that it is possible to reconstruct who generated what.

A common pitfall is treating access as a binary decision: everyone or no one. A more practical approach is to differentiate access by role and risk level instead of opening or closing a tool for the entire organization at once.

Summary

The Goldman Sachs case shows that enthusiasm for AI and risk control are not mutually exclusive, but they require the right sequence of actions. Policies and roles should be defined first, and access expanded afterwards. The reverse order usually ends in withdrawing tools, which costs trust and implementation time.


Lub System helps B2B companies implement AI, automation and IT solutions end-to-end - from strategy to deployment. See our services or get in touch to discuss your case.

Source: https://the-decoder.com/goldman-sachs-pulls-ai-coding-tools-from-traders-and-bankers/