This past month, Goldman Sachs limited the use of artificial intelligence coding tools in their company to only engineers. The company explains that these limitations extend to restrictions on Claude Code and GitHub, two platforms used by bankers to synthesize and analyze information.
The pursuit of restricting AI coding tools at Goldman Sachs grew from an initiative to optimize productivity. Engineers are a key function of the company, providing their knowledge to maintain cybersecurity and software.
However, in addition to engineers, the use of “vibe coding” by traders and bankers without a technical background has raised security and productivity risks because AI-generated code often includes hallucinations, security vulnerabilities and inefficiencies.
Despite these new limitations, however, Goldman Sachs will continue the use of its proprietary AI chat bot “The GS AI Assistant,” which is available to all workers. Nick Carcaterra, a Goldman Sachs spokesperson, said, “After a period of experimentation where we [assessed] 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.”
Other Wall Street firms, including JPMorgan Chase, also expressed their desire to regulate internal AI use. The bank stated that it wanted its employees to explore the best model to use for certain tasks, aiming to avoid the use of expensive models for simple tasks such as analyst reports and save them for more difficult assignments. The bank also placed a strong emphasis on risk management. Lori Beer, the Chief Information Officer at JPMorgan Chase, said, “We want to move at speed, but sometimes we must pull back to get controls in place, the technology is running ahead of capabilities to manage it.”
Bank of America also expressed similar concerns about AI lacking proper human oversight, with CEO Brian Moynihan urging employees to meticulously check AI-generated pitchbooks and spreadsheets.
Goldman Sachs has also used a new holistic approach with the creation of hybrid roles, synchronizing humans with AI systems and emphasizing that AI agents should be employed like humans. “Companies will reskill human managers to oversee a hybrid workforce. The role of human resources will evolve into a department for human and machine resources. The first AI ‘layoffs’ could eventually emerge, in which AI models will be replaced by better AI tools or humans if they perform poorly compared to their peers”, the bank said in a statement.
