Goldman Sachs pulls Claude Code and GitHub from bankers and traders, engineers only
Months after its CIO praised 'vibe coding' by non-technical staff, the bank limited advanced AI coding tools to engineers, citing productivity and risk.
TL;DR
- Goldman Sachs limited Anthropic's Claude Code and Microsoft's GitHub to engineers, according to internal guidance updated in October and reported by Bloomberg on Nov. 2, 2026.
- Traders had used the tools to monitor client trading flows, model market changes and positions, and create written and graphic content.
- Goldman says it reviewed productivity and risk management after a period of experimentation; so-called strats are exempt.
- In July, CIO Marco Argenti had touted vibe coding by traders and bankers whose creations went to the bank's 12,000 engineers.
Wall Street's AI experiment just hit a wall at one of its biggest names. Goldman Sachs has banned most of its staff from using advanced AI coding tools, limiting Anthropic's Claude Code and Microsoft's GitHub to engineers only, according to internal guidance updated in October and seen by Bloomberg News, which reported it on Nov. 2, 2026.
For months, the bank had let bankers use those tools. The ban has frustrated some traders, who used them to monitor client trading flows, model market changes and positions, and create written and graphic content, according to a person familiar with the matter cited by Bloomberg.
From showcase to shutdown
As recently as July, Goldman chief information officer Marco Argenti touted the "vibe coding" of non-technical employees such as traders and bankers. They built solutions with AI tools and handed them to one of the firm's 12,000 engineers to improve, he said in an interview cited by Bloomberg.
"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," Goldman spokesperson Nick Carcaterra said, according to Bloomberg and PYMNTS.
The directive does not apply to so-called strats, employees at the intersection of trading, quantitative research and software engineering. Earlier in 2026, Goldman had already restricted the use of Anthropic's Claude for its staff in Hong Kong, Bloomberg reported.
Banks draw lines around AI
Goldman is not alone in tightening rules. PYMNTS noted that JPMorgan Chase CEO Jamie Dimon warned of "rogue agents" and said outside AI agents would not be permitted inside the bank, while Bank of America and Citigroup set limits on third-party agents. At the same time, Goldman has developed its own AI assistant combining models from OpenAI, Anthropic and Alphabet, and it is lead underwriter on Anthropic's IPO, Bloomberg reported.
What it means for tech and Web3 teams – Analysis
The message from a top bank is that AI coding power is being re-centralized: tools stay, but only with people who can review and own the code. For startups selling AI to regulated finance, governance, audit trails and role-based access now matter as much as model quality. Expect more enterprises to move from "everyone experiments" to "specialists ship".
FAQ
Which AI tools did Goldman Sachs restrict?
Anthropic's Claude Code and Microsoft's GitHub, now limited to engineers, according to Bloomberg.
Why did Goldman restrict them?
The bank said it assessed factors including productivity and risk management after a period of experimentation.
Who is exempt?
So-called strats, who combine trading, quantitative research and software engineering, according to Bloomberg.
When did the change happen?
Internal guidance was updated in October 2026; Bloomberg reported it on Nov. 2, 2026.
Sources
News summary based on the sources above. Not investment, legal or tax advice. Editorial policy & corrections · Report an error
Building in Web3? Turn the news into trust.
Positioning, KOL programs, LinkedIn systems and claim-safe content for Web3 and fintech teams.
Related
OpenAI launches GPT-6 Cyber, its first 'Critical' cyber model, for vetted red teams only
The model scores 48.5% on ExploitGym, found 81 high-severity Firefox bugs and, OpenAI says, stayed in scope in 99.93% of honeypot tests. Access is limited to approved Daybreak Red customers.
OpenAI pledges to pay full power costs and drop NDAs with officials after voters reject data centers
Two days after voters backed data center restrictions in 40 of 41 local ballot measures, OpenAI published a National Community Compact and asked lawmakers to turn similar rules into law.
OpenAI buys Pierre, Meta buys Conductor: big labs absorb AI coding tools in one day
Both deals were announced on Nov. 4, 2026. Pierre's Code Storage will fold into Codex; Conductor stops sign-ups and deletes cloud data from Nov. 20, 2026.