Goldman sachs restricts AI tool use for traders and bankers amid internal dissatisfaction

Goldman Sachs has restricted the use of advanced, code-oriented AI tools by most employees after an internal review assessing efficiency and risk. This move has caused some employee discontent. Previously, the firm permitted bank staff to utilize Anthropic PBC's Claude Code and Microsoft's GitHub for work. According to an internal guideline updated in October, these specific tools are now limited exclusively to engineers. This restriction has frustrated some traders who previously used the tools to expedite tasks such as monitoring client transaction flows, building market volatility models, and generating reports. This policy shift follows earlier praise from Marco Argenti, Goldman Sachs' Chief Information Officer, in July, when he commended non-technical staff like traders for engaging in casual programming using AI to draft initial concepts for refinement by the firm's 12,000 engineers. A spokesperson, Nick Carcaterra, stated the decision was made after evaluating efficiency and risk across various business segments. The firm's AI strategy is part of a broader industry trend; Goldman Sachs itself has developed an internal AI assistant integrating models from OpenAI, Anthropic, and Alphabet. Furthermore, the bank recently acted as the lead underwriter for Anthropic's IPO and partnered with Nvidia on major bond financing deals.