Goldman Sachs pulls AI coding tools from traders and bankers
What happened
Goldman Sachs has removed AI coding tools such as Claude Code and GitHub Copilot from its traders and bankers, restricting access solely to engineering teams. This marks a reversal from earlier this year when CIO Marco Argenti endorsed “vibe coding,” encouraging non-technical staff to use these AI tools. The bank now cites concerns over productivity and risk management as reasons for tightening access.
Why it matters
Limiting AI coding tools to engineers only changes who shapes internal workflows and automation. Traders and bankers previously had the ability to rapidly prototype or tweak code to test strategies or analyze data. Taking these tools away raises operational friction and slows down hands-on experimentation at the front lines. It also signals increased caution around AI risks, such as code errors or uncontrolled outputs, that can disrupt crucial financial functions. Goldman Sachs tightening control on AI coding reflects a trade-off between democratizing AI productivity and mitigating potential liabilities in high-stakes environments.
What to watch next
Watch for whether other financial institutions follow Goldman Sachs in restricting AI tools to technical teams, or if they double down on broader adoption among non-engineers. This move may impact AI governance strategies and tool rollout policies across regulated industries balancing innovation and risk. Also track how Goldman adjusts productivity metrics now that fewer teams have direct AI coding access, and if new oversight processes emerge around AI-generated code. The unfolding approach to AI coding democratization versus control will influence enterprise AI adoption patterns broadly.
AI Quick Briefs Editorial Desk