Goldman Sachs restricts traders and bankers from using certain AI programming tools

According to Bloomberg, Goldman Sachs has restricted most employees from using advanced AI tools aimed at programming after evaluating production efficiency and risk management, leading to dissatisfaction among some employees. Internal guidelines were updated in October, and bankers had previously been able to use Anthropic's Claude Code and Microsoft's GitHub for several months, but the scope of these tools has now been narrowed to engineers.

Sources say that some traders had used these tools to accelerate monitoring of client trading flows, modeling market changes and positions, as well as creating text and graphic content, and this restriction has left them feeling dissatisfied. Goldman Sachs' Chief Information Officer Marco Argenti stated in July that non-technical personnel such as traders and bankers had developed solutions using AI tools, which were then handed over to the company's approximately 12,000 engineers for improvement.

Goldman Sachs spokesperson Nick Carcaterra stated that the company provides employees with certain AI tools to determine applicable scenarios in different businesses; after evaluating factors such as production efficiency and risk management during the trial period, the tools will be given to the most suitable personnel for long-term use. Additionally, sources indicated that the latest directive does not apply to strats who are simultaneously engaged in trading, quantitative research, and software engineering. Earlier this year, Goldman Sachs also restricted employees in Hong Kong from using Anthropic's Claude.