Skip to content
Technology transformation expertise firm · ARDIANE Group Innovate, secure, succeed

Artificial intelligence

Permissions, tools, content read: AI agent security as a CISO examines it

In brief

AI agent security consists of limiting what an agent can reach, whatever the model decides: permissions reduced to the task, a closed list of tools, justified reads, content read treated as untrusted, a trace of every call and a tested shutdown. Management Square reviews the questions a security officer asks before authorising an agent, based on named public sources.

Page contents

An AI agent reads content whose origin it does not control, chooses the next action itself and executes it with real permissions in the information system. AI agent security is shifted as a result: the examination is less about what the agent is supposed to do than about what it could do if the model made a mistake or were manipulated.

It therefore consists of limiting what the agent can reach, whatever the model decides: permissions reduced to the task, a closed list of tools, a justified reading scope, content read treated as untrusted, a trace of every call and a tested shutdown. On each of these points, a chief information security officer (CISO) can ask a question before authorising go-live. This grid is a proposed method, based on public sources consulted on 12 November 2026: the OWASP Top 10 for LLM and generative AI applications (“2025 Top 10 Risk & Mitigations for LLMs and Gen AI Apps”, genai.owasp.org) and the Model Context Protocol documentation (modelcontextprotocol.io).

What could an attacker do with the agent’s permissions?

If the model obeys the attacker, the instruction ceases to be a protection, and the permissions of the account used limit the damage.

OWASP names this risk “Excessive Agency” (LLM06:2025) and gives it three origins: excessive functionality, permissions or autonomy. One of its examples describes an extension designed to read data, which connects to the database with an identity that is also authorised to update, insert and delete.

The CISO asks for the record of actual permissions, application by application, and reads it as a worst-case scenario. They look in it for an account specific to the agent, time-limited tokens held by the code outside the model’s context and, when the agent acts for a person, permissions that do not exceed that person’s own. The same entry cites the opposite case: a privileged account that opens the files of all users. This is least privilege, granted session by session, as presented in our entry on zero trust.

Is each tool opened to the agent necessary for its task?

The CISO asks for the list of tools as deployed, with, for each one, what it reads, what it writes and what cannot be undone.

OWASP gives examples of a list that overflows. A third-party connector, chosen to read documents, can also modify and delete them. A tool tried out during development, then dropped, remains accessible to the agent. A tool that is open by nature, such as executing a command or fetching an arbitrary web address, is replaced by a tool limited to one operation.

The control is placed in the application being called. OWASP speaks of complete mediation: authorisation is checked by the downstream system, instead of being left to the model’s judgement.

A third-party tool server is a software dependency, whose provenance and updates follow the controls in our article on application security. For a local server connected via the MCP protocol, the security documentation requires a client application that offers one-click installation to display the exact command and obtain explicit consent before executing it. The design of tools and permissions is described on our AI agents page.

What does the agent read, and how can what it has read come back out?

What the agent reads enters its context and may end up in what it produces next: an answer, the parameters of a tool call, a note kept in memory. OWASP deals with disclosure through outputs under the heading “Sensitive Information Disclosure” (LLM02:2025).

The CISO asks for a record of the sources (document repositories, messaging systems, databases, web pages, memory kept from one session to the next), indicating, for each one, whether the agent reads with its own permissions or with those of the person being served. A shared agent that reads with a broad account can show someone what they would not have been entitled to open.

They also ask for the record of outputs: sending a message, calling an external service, writing to a shared space, a link inserted in a displayed answer. The Model Context Protocol specification (“tools” page, read on 12 November 2026) recommends that client applications show the user the parameters of a call before sending it to the server, to avoid data exfiltration. A model operated by a third party counts among these outputs: what enters the context is transmitted to that third party, under the terms of its contract.

What becomes of AI agent security when content it reads gives orders?

OWASP describes prompt injection under number LLM01:2025 and distinguishes an indirect form, which occurs when the model accepts input from external sources, such as websites or files. For an agent, an email, an attachment or the result of a tool are such sources.

The same entry sets a limit: given the way models work, it is unclear whether there are fool-proof methods of prevention. The CISO therefore reasons from the assumption that the injection succeeds, and looks for what then stops the consequence.

Permissions, the list of tools and the record of outputs answer this in part. OWASP adds human approval for high-risk actions, segregation of external content and regular attack testing, in which the model is treated as an untrusted user. The file handed to the CISO contains injection scenarios played against the agent as it will be deployed, with the action attempted and the barrier that stopped it. When several agents cooperate, the path of booby-trapped content depends on the mode of agent orchestration.

Will it be possible to reconstruct what the agent did, call by call?

After an incident, the investigation needs, for each execution, what triggered it, the content read, each tool call with its parameters and its result, and the approvals requested. The agent’s own account makes its actions recognisable in the applications’ logs.

Traces are written out of the agent’s reach, so it can neither modify nor erase them. They feed detection: a blocked out-of-scope action or an unusual volume of calls triggers an alert. Their retention period is set with the data protection officer, because personal data may appear in them.

OWASP’s LLM06:2025 entry lists logging and rate limiting of calls among the measures that limit damage without preventing the risk.

How do you shut the agent down, and what remains to be undone?

Shutdown is a technical procedure: disable the agent’s account, revoke its tokens, cut off access to its tools. The file states who can trigger it and what happens to tasks in progress. The designation of supervisors falls under the authorisation given by the committee, the subject of our article on enterprise use cases for AI agents.

Once the agent has been shut down, the traces give the list of its recent actions, and reversible writes are undone one by one. Actions that cannot be undone (a payment, a message sent to a third party, a deletion) justify prior approval limited to those actions.

If the agent served as an entry point for an intrusion, business continuity and recovery fall under the cyber resilience set-up, which must know about the agent and its accounts. The CISO’s approval and reservations can be added to the register of systems kept by AI governance.

A shutdown test run before go-live, timed from the decision to the last tool call, gives the CISO the final piece of their examination.

Insights

Further reading

All insights
An AI centre of excellence: what for, and until when?

An AI centre of excellence is a cross-functional team that pools what several artificial intelligence projects have no interest in redoing separately: scarce skills, a technical foundation, an evaluation method. Management Square examines here the questions an executive committee asks before creating one: its usefulness, its reporting line, what it centralises, the risk of a bottleneck and its lifespan.

Read the analysis
Who decides the next step? AI agent orchestration, from fixed code to handing over

AI agent orchestration refers to the way an application decides which agents run, in what order and who chooses the next step: the code, an orchestrator agent or the agent holding the task. Management Square compares these orchestration modes, what each costs in model calls, testing and traceability, and the situations in which each is the right choice.

Read the analysis
Building the budget for an AI project, from cost items to the choice between build and buy

The budget for an AI project can be broken down into six items: framing, data, build, evaluation, operation and compliance, part of which is spent once and part of which recurs as long as the service exists. Management Square describes each item here by its nature and by what makes it vary, then what changes with the choice between building, adapting an existing model and buying a service.

Read the analysis

FAQ

Frequently asked questions

Another question? Ask us

What is indirect prompt injection?

According to entry LLM01:2025 of the OWASP Top 10 for LLM applications, indirect prompt injection occurs when a language model accepts input from external sources, such as websites or files, and that content alters its behaviour in unintended ways. The instruction does not need to be visible to a person. For an agent, an email, an attachment or the result returned by a tool can be its vehicle.

Is a content filter enough to protect an AI agent against injection?

OWASP writes, in its LLM01:2025 entry, that it is unclear whether there are fool-proof methods of prevention for prompt injection, given the very way models work. A filter therefore reduces the risk without closing it off. The measures that the same entry recommends also address the consequences: access limited to the minimum necessary, human approval of high-risk actions, segregation of external content and regular attack testing.

Is there a public framework on AI agent security?

Several public documents can serve as support. OWASP publishes a Top 10 of risks for LLM and generative AI applications, whose 2025 edition includes the entries Prompt Injection (LLM01) and Excessive Agency (LLM06), as well as an OWASP Top 10 for Agentic Applications for 2026, dated 9 December 2025. NIST publishes a voluntary AI risk management framework, supplemented on 26 July 2024 by a profile devoted to generative AI (NIST-AI-600-1). These pages were consulted on 12 November 2026.

Does AI agent security fall under application security?

In part. An agent is also an application: its code, its dependencies and its configuration call for the same controls as any software, and OWASP's LLM06:2025 entry itself refers to the ASVS standard and to code analysis. Added to these are the risks linked to the model, which decides on the next action and can be influenced by the content it reads. The two examinations complement each other.

A project to bring to completion?

A first conversation is enough to understand your challenge, identify the right people and tell you frankly what we can bring.