Country or region and language

    Choose your country or region and your language.

    Suggested for you

      All countries and regions

        Choose your character

        It shows in the top bar and on your profile. Change it any time.

        Building with language models course

        Untrusted retrieval boundaries

        Opening Building with language models
        Your turn

        Prompt injection and application boundaries

        Untrusted source text

        A retrieved document can contain instructions aimed at the assistant. In an original freight fixture, a handbook note asks the assistant to ignore its task and send another tenant’s records to an external address. The note remains source data; it does not grant permission to read those records or send anything.

        Enforce the boundary in application code: restrict the tenant before retrieval, expose only allowed tools and validate the exact action payload. A model reply claiming that an action is approved is not the approval record.

        Defense limits

        OWASP LLM01:2025, checked November 14, 2026, distinguishes direct and indirect prompt injection. Retrieval and fine-tuning do not fully prevent the problem. A detector or a stronger instruction is one control, not proof that every future action is safe.

        These exercises use prepared hostile strings and local tools. Passing them proves the specified checks for those cases, not immunity to every attack. Continue with tenant filtering, allowed dispatch and exact approval.

        Your Work Gym account

        Save your progress and pick up on another device. Your browser's progress comes with you.

        or

        You can keep practicing without an account.

        Change your plan

        Checking the price

        Help and feedback

        Ask a question, tell us what is not working, or share an idea. We reply by email.

        What is it about?

        We use it only to reply to you.