Goldman restringe a su plantilla el uso de herramientas de IA

Goldman Sachs has sparked discontent among its staff after implementing a policy restricting most employees from using advanced artificial intelligence (AI) tools designed for programming, following an evaluation of productivity and risk concerns. For months, the bank permitted its bankers to use Claude Code from Anthropic and GitHub from Microsoft to carry out their tasks.

However, the financial institution is now limiting the use of these tools exclusively to engineers, according to internal guidelines updated in October, which Bloomberg has accessed.

The prohibition has frustrated some traders who relied on these tools to streamline various aspects of their work, such as tracking client trade flows, creating models on market changes and financial positions, as well as generating written and graphic content, according to informed sources. In July, Goldman's Chief Information Officer Marco Argenti highlighted the phenomenon of "vibe coding" (programming with AI) among non-technical employees, such as operators and bankers.

They had developed solutions using AI tools and subsequently transferred their creations to one of the 12,000 engineers in the company to refine them. "We make certain AI tools available to our employees to identify the most suitable use cases in the various areas of the company," Nick Carcaterra, Goldman's spokesperson, explained.

Banks are investing millions of dollars in AI tools while simultaneously trying to find ways to manage the growing concerns related to cybersecurity risks, costs, and productivity.