Goldman Sachs restricts traders and bankers from using AI coding tools amid risk concerns

Goldman Sachs has restricted most of its traders and bankers from using advanced artificial intelligence (AI) coding tools, including Anthropic's Claude Code and Microsoft's GitHub, following an internal review of productivity and risk management, according to a Bloomberg report.

The Wall Street bank had allowed employees across its business to experiment with the tools for several months but has now limited access primarily to engineers, according to internal guidance updated in October and seen by Bloomberg. The move has reportedly frustrated some traders who had incorporated AI coding assistants into their daily workflows.

Employees had been using the tools to monitor client trading flows, model potential market movements and changes in positions, and generate written and visual content. The restriction does not extend to so-called strats, employees whose roles combine trading, quantitative research and software engineering.

Nick Carcaterra, a Goldman spokesperson, said: “We make certain AI tools available to our people to determine the best use cases across different parts of the firm.”

“After a period of experimentation where we assess a variety of factors including productivity and risk management, we put them in the hands of those who are best suited to leverage them for the long term,” he added.

The decision marks a shift from the bank's earlier encouragement of AI experimentation among non-technical employees. In July, chief information officer Marco Argenti had praised the use of 'vibe coding' by traders and bankers, who were using AI tools to develop applications and solutions before handing them over to engineers for further development.

Argenti had noted that the bank's engineering workforce of approximately 12,000 could refine and improve applications developed by employees without specialist technical expertise.

However, the growing use of AI has also prompted concerns about its implications for employees' analytical and technical capabilities. In August, Goldman partner Chris Churchman warned that increasing reliance on AI could weaken the analytical skills of future generations of bankers, according to Quartz.

Goldman's decision comes as financial institutions weigh the productivity benefits of generative AI against cybersecurity, operational and regulatory risks. Banks are increasingly deploying AI tools across business functions while reassessing access controls, technology costs and the reliability of AI-generated outputs.

JPMorgan Chase said in July that it wanted employees to select AI models based on the complexity of individual tasks, rather than using expensive models for routine activities such as summarising analyst reports.

Goldman has also developed its own AI assistant, combining models from OpenAI, Anthropic and Alphabet. The bank previously restricted access to Anthropic's Claude among employees in Hong Kong, introducing location-specific controls earlier this year.

In May, Goldman's president and chief operating officer John Waldron described the bank as a “human assembly line” with significant potential for automation.

The latest restrictions highlight the challenge facing financial institutions as they seek to expand AI adoption while retaining oversight of how the technology is used, particularly in functions involving sensitive client information and financial operations.