IT Home On November 3, according to Bloomberg, after evaluating efficiency and risk issues, Goldman Sachs banned most employees from using advanced artificial intelligence tools for programming. This measure caused dissatisfaction among some employees.
In recent months, Goldman Sachs has allowed bank employees to use Anthropic PBC’s Claude Code and Microsoft’s GitHub during work. Bloomberg News reported an internal guideline updated in October, which now stipulates that these tools are available only for engineers.
According to a person with knowledge of the matter, this restriction order has caused some traders to feel frustrated. They used these tools to speed up various tasks, such as monitoring customer transaction flows, building models for market fluctuations and positions, and generating text and chart materials.
In July this year, Marco Argenti, CEO of Goldman Sachs’s Chief Information Officer, also praised “casual programming” carried out by non-technical employees such as traders and bankers. In an interview, he said that these employees use artificial intelligence tools to develop initial solutions, which are then handed over to the company’s 12,000 engineers for further optimization.
S Goldman Sachs spokesperson Nick Carcaterra said: “We made some artificial intelligence tools available to employees in order to identify the most suitable applications for different business segments of the group. After a trial period, we evaluated various factors such as efficiency and risk management, and ultimately made the tools available to those who are best suited to utilize them over the long term.”
Wall Street is still trying to figure out how to standardize the use of artificial intelligence models among thousands of employees, and there are new developments in this regard. Major banks are investing millions of dollars in artificial intelligence tools, while also dealing with increasingly prominent cybersecurity risks, costs, and efficiency issues.
IT Home noticed that JPMorgan Chase said in July that it is guiding employees to choose the right model according to needs, avoiding simple tasks such as generating analyst report summaries, and also avoiding using the most costly large models. Goldman Sachs has also developed an artificial intelligence assistant that integrates multiple models from OpenAI, Anthropic, and Alphabet.
At the same time, Wall Street institutions are also benefiting from the surge in financing needs for artificial intelligence companies. Goldman Sachs served as the lead underwriter for Anthropic’s major initial public offering, and it cooperated with chip manufacturer NVIDIA to arrange several large-scale bond financings.
In May this year, John Waldron, who is seen as the successor to CEO David Solomon, said that Goldman Sachs is like an “artificial assembly line” that is very suitable for introducing automation.
This is not the first time Goldman Sachs has restricted employees’ use of artificial intelligence tools. Earlier this year, the bank introduced separate restrictions for employees in Hong Kong, prohibiting local staff from using Claude from Anthropic.
Another insider revealed that Goldman Sachs’ latest directive does not apply to employees in quantitative strategy positions, who also handle trading, quantitative research, and software development tasks.