Goldman Sachs restricts AI coding tool access to developers

Goldman Sachs has implemented new internal guidelines that restrict the use of advanced AI coding tools, including Anthropic’s Claude Code and Microsoft’s GitHub, primarily to its engineering staff. This policy, effective as of October, marks a shift from a previous experimental phase where traders and bankers were permitted to use these tools for tasks such as monitoring client trade flows, modeling market changes, and creating content.

The restriction has reportedly caused dissatisfaction among some traders who utilized the technology to accelerate their workflows. However, the bank’s spokesperson, Nick Carcaterra, stated that following a trial period, the tools are being returned to those “best suited to use them long-term” to ensure better monitoring of productivity and risk management.

The new rules do not apply to “strats,” employees whose roles sit at the intersection of trading, quantitative research, and software engineering. This move follows earlier, more localized restrictions, such as those placed on employees in Hong Kong.