Investigation, verification, protection, and vetting — built as one connected infrastructure for institutions that interact with children, survivors, and people in crisis.
The federal government's own accountability tools are disappearing. Meta ended its fact-checking program. OpenSecrets laid off a third of its staff. The National Law Enforcement Accountability Database was shut down. The infrastructure that once held institutions to account is being dismantled — at exactly the moment AI-generated misinformation is accelerating beyond anyone's ability to fact-check it by hand.
Integrity Alliance is a coordinated response. Seven products, one mission: make corporate, governmental, and AI-generated conduct legible to the public — with primary sources, published methodology, and editorial independence written into the governance itself.
We do not censor. We document. We do not adjudicate guilt or innocence. We let primary sources speak. And we publish our standards before we publish a single story, so any reader can audit our work the way we audit everyone else's.
Each product solves a specific accountability gap. Together they form connected infrastructure — investigation, verification, protection, and vetting — under a single editorial standard.
Independent. Investigative. Uncompromising.
The investigative publication of record. Makes corporate, financial, and institutional conduct legible to the public — with primary-source verification, four-persona editorial audit, legal review, and subject comment before any story publishes. Inaugural investigations include the Bonnie Plants environmental and data opacity record and the Lifetouch student-photography pipeline.
Is it real, and is it human?
The synthetic-media and claim-verification arm. Identifies AI-generated content, scores authorial trustworthiness against primary sources, and flags coordinated amplification patterns. Discern reports are the fact-check layer beneath every Watchdog investigation — and a standalone publishing format for emerging misinformation events that need fast, defensible response.
Protected-identity data infrastructure for populations the existing systems fail.
A cryptographic standard for handling identity data about people who cannot afford to be identified — minors in outcome research, domestic-violence survivors, trafficking witnesses, undocumented workers reporting employer misconduct. One-way HMAC tokenization with Shamir split-key custody and append-only Merkle audit logs. Built so no single person — including the founder — can re-identify a subject from internal data access alone.
Independent vetting for entities, facilities, and AI personas with access to children.
A six-dimension vetting framework applied identically to public, private, and government entities — and to AI personas — that seek access to children. Identity verification, criminal and registry records, behavioral attestation, facility inspection, funding and ownership disclosure, and AI-persona provenance. Findings published with primary-source evidence using the Verified / Caution / Avoid taxonomy. No entity is automatically trusted.
Forensic environmental psychology — wellbeing through every business atmosphere.
A sister product that measures how color, light, sound, and layout flow through any physical space — schools, daycares, clinics, factories, restaurants, offices. Operators use it to score and remediate their environments across four pillars: safety, health, welcoming atmosphere, and revenue. Applied with discipline, the same framework documents what corporate facades conceal from regulators — making it both an operator self-improvement tool and an investigative methodology.
A trust score for online claims.
A Carfax-style report card for online articles and the people who write them. Extracts every factual claim from a piece, checks each against primary government sources — IRS filings, court documents, congressional votes, inspector general reports — and generates an authorial trust score from 0 to 100. Demonstrated on a real-world case: a pseudonymous author's viral piece accusing a legitimate children's legal-aid organization of trafficking. Of eight claims: four fabricated, two twisted out of context, two true. Author score: 18 / 100.
Is this written by a human?
A Chrome extension that scans any webpage and tells you whether the content was AI-generated or written by a real person. Maintains a verified Trust Registry of 500+ confirmed human journalists and publications. Built on GPTZero's detection technology. Together with TruthMark, it answers both halves of the question every reader needs to ask before they share a story: is it true, and is it human?
The seven products are not separate ventures. They are one connected infrastructure with a deliberate division of labor:
Ambience Integrity provides the field methodology for any physical-space investigation. TruthMark and TruthLens are the public-facing tools through which Discern's underlying engine becomes accessible to readers, parents, journalists, and platforms.
The discipline is the same across every product: primary sources, published methodology, independent editorial review, and governance written so that the founder cannot quietly compromise it later. The Editorial Independence Charter survives organizational changes including dissolution, sale, or merger.
We're seeking founding funders, organizational partners in child welfare and accountability journalism, and technical talent who want to build something that matters. Reach out — or download the full product briefs for your team.