Skip to content

Questions and answers

AI agent security questions, answered

The short answer

These are the AI agent security questions owners ask us most, with short, direct answers. Each one links to the page that goes deeper. If yours isn't here, book a walkthrough and we'll answer it on your own work, with your own tools in front of us.

Getting started

What is an AI agent?

An AI agent is software that uses an AI model to decide on and carry out steps with tools, like reading email, updating a CRM or drafting a reply, within the access it has been given.

Is it safe to use AI agents in a small business?

Yes, when the system around them is built for it: each agent has its own identity, only the access its job needs, a person approves anything that leaves the business, and every action is logged.

What should an AI agent do first?

Start with low-risk internal work like drafting replies, summarizing threads or preparing reports. Expand to actions that reach customers once you've seen how it behaves.

Do I need a security team to run AI agents?

No. Most of the protection comes from choosing a system with approvals, scoped access and logging built in. Those are decisions made once, not a team watching all day.

Control and approvals

Can an AI agent send email without me?

It shouldn't, unless you've given a standing approval for that kind of email. By default an agent should draft and wait for the person who owns the send.

What actions should always need a person's approval?

Sending email or messages, posting, publishing, paying or spending, deleting, and changing permissions, credentials or a live website.

What is a standing approval?

A standing approval lets an agent do one kind of routine work without asking each time. You give it after you trust that work, and you can take it back at any time.

Who is responsible when an AI agent makes a mistake?

The business is, which is why every high-impact action should name the person who approved it. A good audit trail shows which agent acted and who said yes.

Risks

What is prompt injection?

Prompt injection is an instruction hidden in content an AI reads, such as an email or web page, meant to make it do something its owner didn't ask for. You contain it by limiting what the agent can do and requiring approval for risky actions.

What is shadow AI?

Shadow AI is AI tools people use for work without the business approving them, often connected to company email or files. An approved tool that's just as easy is the most reliable fix.

What happens if an AI agent's key leaks?

Whoever holds the key can act as the agent. Keys should live in a vault, be scoped to one job, rotate regularly, and be easy to revoke in one place.

What is the OWASP Top 10 for Agentic Applications?

It's OWASP's list of the ten most important risks for AI agents, published December 9, 2025, from agent goal hijack (ASI01) to rogue agents (ASI10).

Rules and Massachusetts

Does 201 CMR 17.00 apply to AI agents?

It applies to any business holding personal information about Massachusetts residents, and it doesn't exempt software. Restricted access, unique IDs and monitoring cover your agents too.

Do I have to report a breach caused by an AI agent?

In Massachusetts, M.G.L. c. 93H requires notice when personal information is acquired or used without authorization, whatever the cause. An audit trail helps you work out what was touched.

What does NIST say about AI agents?

NIST's Cybersecurity Framework 2.0 covers services alongside users and asks for least privilege and separation of duties (PR.AA-05). Its AI Risk Management Framework asks for clear accountability and human oversight.

Where can I see how LTFI handles this?

Book a walkthrough and we'll show you the approval queue, standing approvals and the activity log on real work.

Keep reading

  • A heavy threshold gate lets one cleared path through while dozens of pink trails stop at it.

    Start here

    Agentic AI security

    Agentic AI security explained for owners: identity, least privilege, approvals and an audit trail for AI agents, from the team that runs them in Massachusetts.

    Brian Gagne

  • Two interlocking rings, a lock dial and an open aperture, joined by an amber clasp.

    Operations

    Agentic IAM and AI SOC in 2026

    Agentic IAM and AI SOC are the two 2026 security predictions that reach a small business. What they mean, what to ask your IT provider, and what to do first.

    Amelia Gagne

  • Medallions tangled into one pink knot, and one free medallion with its own amber thread.

    Identity and access

    Agentic IAM

    Agentic IAM gives every AI agent its own identity, scoped short-lived credentials and authority it can trace to a person. How it works and where to start.

    Brian Gagne

  • A pink stream of light is held at a round amber valve gate, waiting for a person's yes before anything leaves.

    Oversight

    Human approval for AI agents

    Human approval for AI agents that people actually read: which actions need a yes, how to avoid approval fatigue, and how standing approvals work.

    Amelia Gagne

  • A blurred hub drowns in pink cables while a clean hub in front holds one amber line.

    Identity and access

    AI agent permissions

    AI agent permissions decide what an agent can read, change and send. How to scope its tools, catch permission drift and run an access review.

    Brian Gagne

  • Hidden pink-lit devices behind dark panels, and one wide amber archway beside them.

    Threats

    Shadow AI

    Shadow AI is the AI your team uses without approval. How to find it without spying, write an AI policy people follow, and offer a better tool.

    Amelia Gagne

  • A sprawling swarm of small pink-lit tags, with one sealed and glowing amber inside a single capsule.

    Identity and access

    Non-human identities

    What a non-human identity is, why AI agents multiply them, and how to vault, scope and rotate API keys, tokens and service accounts before one leaks.

    Brian Gagne

  • A capsule travels through a line of rings and exits cleanly through an amber one.

    Identity and access

    AI agent lifecycle

    The AI agent lifecycle, run like staffing: who requests and owns an agent, what it gets on day one, how access is reviewed, and how to retire it cleanly.

    Amelia Gagne

  • Many pink cables converge on one hub, and every connection passes through its amber iris.

    Threats

    MCP security

    MCP security for businesses running AI agents: what the Model Context Protocol's authorization rules require, and how to vet servers and stop tool misuse.

    Brian Gagne

  • A column of stacked plates with pink-lit seams sealed by an amber top.

    Oversight

    Audit trails for AI agents

    An AI agent audit log proves what an agent did, who asked and who approved it. What to record, how to keep it tamper evident, and how long to keep it.

    Amelia Gagne

  • A smoked translucent slab carries a hidden pink thread toward the agent's tools, and an amber ring stops it before it reaches them.

    Threats

    Prompt injection

    Prompt injection turns an email, a web page or a file into orders for your AI agents. Why it can't be trained away, and the controls that limit the damage.

    Brian Gagne

  • Pink beads of personal detail drift between dark vessels, and one sealed amber vial holds only a few.

    Operations

    AI agents and personal data

    AI agents and personal data: what an agent should see, what it may keep, how to delete it on request, and how to log its work without hoarding personal details.

    Amelia Gagne

  • A swarm of pink alert shards streams through a dark sorting lens, while one amber lever stays locked until a person decides.

    Operations

    The AI SOC

    An AI SOC lets agents triage, enrich and summarize alerts. Containment, lockouts, deleting data and closing incidents still need a person to approve them.

    Brian Gagne

  • A column of vault tumblers lights pink at each seam, and the last aligns amber as the vault door edge begins to glow.

    Oversight

    AI agent security checklist

    An AI agent security checklist for owners: what to check before an agent acts, then identity, approvals, data, logging, incidents and retirement.

    Amelia Gagne

  • Ten black pillars veined with pink fractures stand in a row, and three amber rings thread through every one of them.

    Threats

    OWASP Agentic Top 10

    The OWASP Agentic Top 10, ASI01 to ASI10, explained for owners: each risk in everyday terms, a small-business example, and the control that answers it.

    Brian Gagne

  • Reference

    Glossary

    An agentic AI glossary in plain words: AI agents, agentic IAM, non-human identities, prompt injection, shadow AI, MCP and more, each with a short definition.

Most MSPs are resellers. Kief Studio is the manufacturer.

A reseller sends your problem to the company that made the software. We are that company: we build LTFI and run it, so the people you talk to can change it. Book a walkthrough and see it on your own business.

Nothing plays until you press play.