TL;DR
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
Goldman Sachs has restricted advanced AI coding tools to engineers after a period of experimentation, according to Finextra’s account of a Bloomberg report. The bank cited productivity and risk management as factors in its review; the effects on individual teams and the duration of the restriction have not been disclosed.
Goldman Sachs has restricted advanced AI coding tools to engineers, limiting access for traders and other employees after a period of testing, according to Finextra’s report citing Bloomberg. The change affects staff who had used tools such as Anthropic’s Claude Code and Microsoft’s GitHub products for coding tasks, and the bank said its review considered productivity and risk management.
Finextra reported that Goldman had previously allowed bankers to experiment with tools including Claude Code and GitHub. Under the reported policy change, use is now limited to engineers. The report does not specify which individual products or versions are covered, or whether every non-engineering employee has lost access.
Some traders were frustrated by the restriction, Bloomberg reported, because they had used the tools for tasks such as monitoring client trading flows and modelling market changes and positions. The report does not quantify how many employees were affected or how much work had been built around those uses.
Goldman spokesperson Nick Carcaterra told Bloomberg that the firm makes certain AI tools available so it can identify their best uses across different parts of the business. He said that after an experimental period, the bank considers factors including productivity and risk management before deciding which employees are best placed to use tools over the longer term.
AI Access Shifts to Engineering Teams
The restriction illustrates a tension facing financial firms as generative AI tools become easier for non-specialists to use: they can help employees build small applications or automate tasks, but work produced outside formal engineering teams may raise questions about quality, oversight and operational risk. Goldman’s stated review factors show that the bank is weighing those considerations alongside potential productivity gains.
For traders, limiting access could affect how quickly they can prototype tools for analysis or desk-level workflows. The report gives examples of uses involving client flows and market modelling, but does not establish that the tools changed trading decisions or produced measurable gains. The policy’s practical effect will depend on whether engineers can support those requests and how the bank handles projects begun during the trial period.
The move also matters beyond Goldman because banks operate in a regulated environment and handle sensitive financial and client information. The report does not say that a breach or specific failure prompted the decision. It describes a policy change following experimentation, not a finding that the tools caused harm.
Popular right now
Vitamix Propel Series 750 Professional-Grade Blender, 4 Programs, Black
Ninja Foodi Air Fryer, 8 Qt DualZone, 6-in-1, Grey, DZ201
Dyson V8 Cyclone Cordless Vacuum
As an affiliate, we earn on qualifying purchases.
Goldman’s Earlier AI Experiments
The restriction follows a period in which Goldman employees were allowed to test AI coding tools. During the summer, Chief Information Officer Marco Argenti discussed the potential of “vibe coding” by non-technical staff, with tools created in that way then passed to engineers, according to the source report. That description suggests experimentation could include employee-built prototypes that were later reviewed by technical teams.
There had also been internal warnings about relying too heavily on AI. In August, Goldman partner Chris Churchman, who leads one of the bank’s flagship AI projects, warned that the technology could weaken bankers’ reasoning skills if employees depended on it excessively, Finextra reported. That warning is a concern about possible effects, not evidence that such degradation occurred.
Finextra also reported that earlier in the year Goldman was among banks that restricted employee access to Claude models in Hong Kong amid growing US-China tensions over AI. That regional restriction is distinct from the coding-tool policy described in the current report; the source does not say the two decisions had the same cause.
““We make certain AI tools available to our people to determine the best use cases across different parts of the firm.””
— Nick Carcaterra, Goldman Sachs spokesperson, speaking to Bloomberg
Policy Scope and Rationale Remain Unclear
The available report does not give the date the restriction took effect, its expected duration or whether it applies uniformly across Goldman’s businesses and locations. It also does not identify exactly which tools are restricted, what access engineers retain, or whether non-engineers can still use approved AI products for tasks that do not involve coding.
Goldman’s spokesperson cited productivity and risk management as factors, but the bank has not disclosed specific test results, incidents or performance measures behind the decision. It is also unclear whether the restriction followed concerns about code quality, data handling, employee reliance on AI, or a combination of issues. The source does not report a particular security incident or loss.
Bloomberg’s account, as summarized by Finextra, says some traders were frustrated. The number of employees affected, the scale of their prior use and whether those workflows will be transferred to engineering teams have not been reported.
How Goldman Manages AI Workflows
The next relevant development will be how Goldman defines approved access and routes requests from traders and other non-engineering staff. The bank has not announced a timetable for reviewing the restriction or said whether access could expand again after further testing.
Employees and clients will also be watching whether engineering teams take responsibility for applications previously prototyped by business staff, and whether Goldman publishes further details about its evaluation process. For now, the confirmed information is limited to the reported access change and the bank’s explanation that it assesses productivity and risk management before assigning longer-term use.
Key Questions
What has Goldman Sachs changed?
According to Finextra’s report citing Bloomberg, Goldman has limited advanced AI coding tools to engineers, restricting access for traders and other employees who had been allowed to experiment with them.
Which AI coding tools are mentioned?
The report names Anthropic’s Claude Code and Microsoft’s GitHub. It does not provide a full list of affected tools or specify which versions are covered.
Why did Goldman restrict access?
Goldman spokesperson Nick Carcaterra told Bloomberg that the firm evaluates potential uses after a period of experimentation, considering factors including productivity and risk management. The bank has not disclosed detailed findings or said that a specific incident triggered the change.
Will Goldman employees still be able to use AI?
The report describes limits on advanced AI tools for coding and says engineers retain access. It does not clarify what other AI tools non-engineers can use, whether exceptions are available, or how the policy applies across teams.
Source: finextra.com
