Detecting Fraud in Net Zero

Humanly partnered with Guardian AI, the compliance and workflow technology behind My Eco Move to support the development of a forensic document and AI detection capability for fraud prevention within the ECO4 domestic retrofit scheme.

ECO4 relies heavily on digital evidence to validate eligibility, property condition, installed measures and compliance with scheme requirements. Images, documents and supporting records are submitted at scale by delivery partners, installers and supply chains to justify funding, demonstrate works completion and satisfy audit requirements. As delivery volumes increase, so does the risk associated with relying on digital submissions alone.

The project was initiated in response to growing concern around evidence integrity within retrofit programmes. Reused imagery, altered photographs, misrepresented property conditions and manipulated documentation all present material risk to scheme administrators, local authorities and funding bodies. Individually, these issues may appear minor. At programme level, they can lead to incorrect funding decisions, audit failures, remedial works and reputational damage.

Humanly’s role in the partnership was to provide an authenticity detection layer focused on property related evidence. Using the Property Claims and Grant Evidence Model, Humanly supported Guardian AI in assessing whether submitted images and documents showed indicators of manipulation, editing or AI generation before being relied upon in compliance workflows.

The integration was designed to complement existing compliance controls rather than replace them. Guardian AI already manages structured workflows across assessment, design, installation and validation stages. Humanly’s detection capability adds an additional assurance step, strengthening confidence in the digital evidence underpinning those processes.

This approach allows higher risk submissions to be flagged for further review without slowing down legitimate delivery. It also supports earlier identification of systemic issues across supply chains, such as repeated image reuse or inconsistent evidence patterns, enabling programme managers to intervene before issues escalate.

The partnership reflects a broader shift within retrofit and property related grant programmes. As evidence submission becomes increasingly digital and distributed, trust in that evidence can no longer be assumed. Combining workflow intelligence with forensic authenticity analysis provides a scalable way to protect public funding, support audit readiness and maintain confidence in scheme delivery.

By embedding Humanly’s detection capability within Guardian AI’s retrofit platform, the project demonstrates how authenticity verification can be applied practically in real world property and grant funded environments, without adding friction for compliant delivery partners.