3. Permissions by design: Bind instruments to duties, to not fashions
A typical anti-pattern is to offer the mannequin a long-lived credential and hope prompts hold it well mannered. SAIF and NIST argue the other: credentials and scopes ought to be certain to instruments and duties, rotated frequently, and auditable. Brokers then request narrowly scoped capabilities by way of these instruments.
In observe, that appears like: “finance-ops-agent might learn, however not write, sure ledgers with out CFO approval.”
The CEO query: Can we revoke a selected functionality from an agent with out re-architecting the entire system?
Management knowledge and habits
These steps gate inputs, outputs, and constrain habits.
4. Inputs, reminiscence, and RAG: Deal with exterior content material as hostile till confirmed in any other case
Most agent incidents begin with sneaky knowledge: a poisoned net web page, PDF, electronic mail, or repository that smuggles adversarial directions into the system. OWASP’s prompt-injection cheat sheet and OpenAI’s personal steerage each insist on strict separation of system directions from consumer content material and on treating unvetted retrieval sources as untrusted.
Operationally, gate earlier than something enters retrieval or long-term reminiscence: new sources are reviewed, tagged, and onboarded; persistent reminiscence is disabled when untrusted context is current; provenance is connected to every chunk.
The CEO query: Can we enumerate each exterior content material supply our brokers study from, and who permitted them?
5. Output dealing with and rendering: Nothing executes “simply because the mannequin stated so”
Within the Anthropic case, AI-generated exploit code and credential dumps flowed straight into motion. Any output that may trigger a facet impact wants a validator between the agent and the actual world. OWASP’s insecure output dealing with class is express on this level, as are browser safety finest practices round origin boundaries.
