Red flags when hiring an AI agency
The clearest red flags when hiring an AI agency are: no verifiable case studies or contactable references, an unwillingness to name the people doing the work, vague data-handling answers, pressure for a large upfront payment outside escrow, undefined scope with no acceptance criteria, reluctance to sign a Statement of Work, no clear answer on who owns the code and IP, and a demo that can't be traced to a real production project. Any one warrants caution; two or more is a reason to walk away.
A failed AI project rarely fails on the technology alone — it fails on the things an agency was vague about before you signed. Each warning sign below maps to a specific way projects go wrong.
No verifiable case studies or references
If every project is 'under NDA' and no client can ever be contacted, you have no way to confirm the agency has shipped anything. Real agencies can point to production work and put you in touch with at least a couple of past clients.
They won't name who does the work
A polished sales engineer is not who builds your project. If the agency won't tell you which named people will be on your team, or their backgrounds don't match your problem class, you are buying a black box.
Vague answers on data handling
Ask where your data lives, who can access it, how long it's retained, and whether it trains models. Vagueness or surprise at the question means data governance is an afterthought — a serious problem under GDPR and the EU AI Act.
Pressure for large upfront payment outside escrow
A deposit is normal; demanding most of the fee upfront with no escrow is not. It shifts all the risk to you and removes your only leverage if delivery slips. Confident agencies are happy to be paid milestone by milestone.
Undefined scope and no acceptance criteria
If nobody has written down what 'done' means for each milestone, disputes are inevitable. Scope defined only by hours or dates — with no testable deliverable — is a setup for overruns.
Reluctance to sign a Statement of Work
A Statement of Work is a basic professional expectation. An agency that resists one, or 'works on trust', is avoiding accountability for scope, timelines, and IP.
No clear answer on IP and code ownership
If they can't tell you in writing that you own the code, the fine-tuned model, and the derivative work, assume you don't. Ownership disputes surface exactly when the project has succeeded and is most valuable.
A demo that can't be traced to production
Anyone can wire up an impressive demo. Ask what of it is live in production, for whom, and at what scale. A demo with no production lineage is marketing, not evidence.
How to protect yourself
Three safeguards neutralise most of these risks: verify the agency is a registered company, insist on a Statement of Work with defined milestones and acceptance criteria, and pay through escrow so funds release only when you approve each milestone. For the full evaluation process, read how to vet an AI agency, and for the exact questions to put to them, see questions to ask an AI agency before you hire.
Removing the red flags entirely
NorthBridge AI exists to take these risks off the table. Every agency is verified — including company registration — before it can list, every project ships with a signed Statement of Work and escrow-protected milestones, and independent mediators step in if a dispute arises. See how it works.
Frequently asked questions
What are red flags when hiring an AI agency?
The main red flags are: no verifiable case studies or contactable references, unwillingness to name the people doing the work, no clear policy on data handling or model training, pressure for large upfront payments outside escrow, vague scope with no defined milestones or acceptance criteria, reluctance to sign a Statement of Work, no answer on who owns the code and IP, and a demo that cannot be traced to a shipped, production project.
How can you tell if an AI agency is a scam?
Warning signs include a company you cannot verify in any business register, insistence on full payment upfront and outside escrow, references that cannot be contacted, no written contract, and pressure to sign quickly. Protect yourself by verifying company registration, using a Statement of Work, and paying in escrow-protected milestones so funds are released only when you approve each deliverable.
Is an upfront payment a red flag when hiring an AI agency?
A modest deposit is normal. A demand for a large share of the total fee upfront and outside any escrow arrangement is a red flag, because it removes your leverage if delivery falls short. Prefer milestone-based payments held in escrow and released only on your approval.
Skip the risk — hire verified agencies
Browse pre-vetted AI agencies on NorthBridge AI and hire with escrow protection.
Browse verified agencies