AI Human Proof™

AI Human Proof™ is an assurance framework operated by 3RD4PR. It examines the human controls behind your AI systems, agents and automated decisions — checking that they remain transparent, accountable, secure and subject to meaningful human control. Systems are assessed against seven control domains and given a maturity level in each.

It means the human controls around an AI deployment were verified at the date of assessment, within an agreed scope, and reached Level 3 or above across all seven domains. Findings and required safeguards are documented. It is an assessment at a point in time, subject to continuing monitoring and review — not an accreditation, a legal opinion, or a guarantee that a system is safe, compliant or incapable of failure.

The Audit tier starts at £6,500 and takes two to three weeks. Deploy — where controls are built and the assessment issued — starts at £22,000 over six to ten weeks. Ongoing Assure starts at £2,400 per month. Ranges are for a single-agent deployment in one environment; multi-agent, multi-market and regulated-sector work is scoped separately. All figures exclude VAT and third-party platform costs.

Your technical team will usually have handled permissions and infrastructure well. The part that is commonly missing is what happens after something goes out: disclosure position, holding statement, who decides, who tells the customer, and whether anyone has actually tried to manipulate the agent on purpose. We work alongside technical teams, not instead of them.

A policy states what should happen. An audit tests what actually does. We attempt to manipulate the agent the way a bad actor would — through hostile webpages, documents, emails and hidden instructions — and report the results. The documentation records what was found and what was changed, rather than what was intended.

Prompt injection is when hidden instructions inside content the agent reads — a webpage, an uploaded CV, a supplier document, an email — redirect its behaviour. It matters because the instruction does not come from your team and does not look like an attack. Testing for it is part of every AI Human Proof™ assessment, and it is repeated whenever the model, prompts or vendor change.

The seven domains are universal, so the framework applies to any business deploying AI. We maintain sector practices for PR and communications, legal, financial services, healthcare and clinical, e-commerce and retail, and recruitment and HR — each with its own risk model, vocabulary and regulatory position. Other sectors are scoped on request. We implement to the applicable standard and bring specialist regulatory, legal or clinical input where required.

Two to three weeks for the Audit tier, from kick-off to the board summary. Deploy runs six to ten weeks depending on how many systems the agent touches and how much remediation the audit finds. Assure is continuous, with formal re-testing and re-assessment each quarter.

That is one of the most common reasons businesses come to us. The Statement of Assessment and board summary are written to be forwarded directly to a customer, insurer or procurement team. Being able to answer the question clearly increasingly decides whether work is won or stalls.

No, and you should be cautious of anyone who does. What the framework provides is that failure becomes far less likely, that you find out faster when something goes wrong, that a rehearsed response exists, and that you hold documented evidence of the reasonable steps you took. That evidence matters considerably when you are explaining yourself afterwards.

With the Audit. If AI is already touching your systems, it is the fastest way to establish what it can actually do — and what it could do on its worst day. Fixed fee, two to three weeks, with no commitment to what follows.