All guides
    AI news
    4 min read

    Goldman Sachs limits AI coding tools to its engineers

    Bloomberg reported on November 2 that Goldman Sachs has restricted AI coding tools to its engineers, citing productivity and risk management. Here is what is known and what to take from it.

    Goldman Sachs limits AI coding tools to its engineers

    On November 2, 2026, Bloomberg reported that Goldman Sachs is limiting its use of AI coding tools. According to the report and the coverage that followed it, internal guidance updated in October restricts Claude Code and GitHub's tools to the bank's engineers, which removes them from most bankers and traders. This post covers only what those outlets reported.

    What was reported

    • Access to the tools is now confined to engineers. Coverage cites a figure of about 12,000 engineers, and says the engineers who also do quantitative work keep access.
    • A Goldman spokesperson, Nick Carcaterra, was quoted as citing "a variety of factors including productivity and risk management." He was also quoted as saying the experimentation phase is over and the tools are going to people best suited to use them over the long term.
    • Traders and bankers had been using the tools for tasks such as monitoring client trading flows, modelling market changes and positions, and producing written and graphic content.
    • The reports say the change has caused frustration inside the firm. They also note that as recently as July the bank's chief information officer spoke publicly in favour of non-technical staff building software with AI.

    The outlets differ slightly in tone: some describe a restriction to engineers, others call it a ban for most staff. The underlying guidance has not been published, and the reports do not say what specific risk led to the decision. The bank has not, in these accounts, described a particular incident.

    Why it matters

    For the last two years the usual story has been that more people at work should use more AI. This is a large employer drawing a line the other way: the tools stay, but fewer people may use them, and the people who do are those trained to check the output. That is a reasonable position, and a common one in regulated work. Code written by an AI tool still has to be reviewed by someone who can read it, and a banker who cannot read code cannot tell when the tool has got it wrong.

    It also shows how quickly workplace rules around AI are still moving. Guidance written in the spring can look different by autumn.

    What it means for families, schools and frum professionals

    • If you work in finance, law, accounting, insurance or any field with client confidentiality, find out what your employer permits before you paste anything into an AI tool. Policies are changing, and an old email may no longer reflect the current rule.
    • If you run a small business or a communal organisation, write down which AI tools staff may use, for what, and who checks the results. A one-page policy is better than none.
    • For parents: the same question applies at home. A tool is only as safe as the person who can judge its output. A child who cannot yet judge whether an answer is right needs a tool with guardrails and an adult nearby.
    • For schools and mechanchim: this is a useful example for students about to enter the workforce. Using AI well includes knowing when you are not the right person to rely on it.

    What it does and does not change for Navōn

    A decision at one bank does not change how Navōn works. Navōn is not a coding tool for banks, and nothing in this report is about it. What Navōn does is described on the features page: messages are checked on the server against the standard that applies to the account, and parents and schools set the rules for a child's account. No filter is perfect, and Navōn does not pasken: halachic questions belong with a rav. The general lesson of the Goldman story, that access to AI should match the person's ability to check the result, is one we would apply to a child's account as much as an employee's.

    See how Navōn sets rules for a child's account.

    Common questions

    Why did Goldman Sachs restrict AI coding tools?

    A Goldman spokesperson was quoted by Bloomberg citing a variety of factors including productivity and risk management. The reports do not name a specific incident.

    Who can still use Claude Code at Goldman Sachs?

    According to the coverage, the tools are confined to the bank's engineers, reported as about 12,000 people, including engineers who do quantitative work.

    Can my employer restrict AI tools?

    Employers set their own rules for the tools used on work systems and with work data. Check your current policy, and ask if it is unclear.

    Is it safe to put client information into an AI chatbot?

    That depends on your employer's rules, your professional obligations and the tool's data terms. Check all three before you do, and ask a rav for any halachic question.

    Sources

    Published by Navōn. How these guides are written and checked.

    Read next