Vol. 1 · No. 1 An Independent Review Updated May 16, 2026
The Marici
Accountability Review

A claim-by-claim record of public statements made by the nonprofit Marici (marici.org), and what could be verified from public sources.

§ For Donors

How to evaluate any nonprofit's AI & impact claims.

The guidance below is not unique to Marici. It is what conscientious donor due-diligence looks like in 2026, for any organisation making AI or large-scale impact claims.

Charity is a sector built on trust. Trust is well-placed when organisations operate transparently and substantiate their claims; it is exploited when they don't. The questions below are designed to help donors distinguish between the two, regardless of the cause.

None of these checks require specialist knowledge. They require five to thirty minutes per organisation and a willingness to look past donor-facing copy.

01

Watch dollar-to-outcome ratios.

“$X saves N lives” claims should be supported by audited cost-per-outcome data. Ask: what does “saves” mean here? Who verifies the outcome? Does the math reconcile with the organisation's published financials? (If they claim 39,000 rescues on a $5M budget, that's $128 each, not the $500 their donor page asks for.)

02

Look for technical fingerprints.

Real AI organisations leave artefacts: engineering hires on LinkedIn, GitHub repositories, technical papers or whitepapers, infrastructure vendor disclosures, named model providers. Marketing copy (“world's most sophisticated AI”) is not evidence of capability. If you cannot find a single engineer's LinkedIn, ask why.

03

Demand named jurisdictions.

“We operate in six cities” with no city named is unverifiable. Real partnerships with law enforcement leave records: joint press releases, named-agency statements, court filings, government registrations. The absence of named cities or agencies is a meaningful absence, not a privacy-preservation choice.

04

Read the Form 990, not the website.

IRS Form 990 filings are public on ProPublica Nonprofit Explorer. They reveal executive compensation, fundraising costs, related-party transactions, and operational scale — the things marketing copy cannot or will not. A two-minute search of ProPublica often answers more than a thirty-minute browse of a charity's website.

05

Check for independent audits.

For organisations above ~$2M in revenue, look for a publicly posted independent audit conducted by a named accounting firm. Its absence is a meaningful signal. Charity Navigator and Candid both track this disclosure; if it's flagged as missing, ask the organisation directly.

06

Beware superlatives.

“World's most sophisticated.” “Most ambitious effort in the modern world.” “First-of-its-kind.” Superlatives in donor-facing copy without third-party attribution should be read as marketing language, not measurement. Real superlatives are awarded by independent bodies, not self-declared.

§ A note on emotional manipulation

The hardest part of giving responsibly.

Causes involving children, trafficking, abuse, and rescue are designed — emotionally — to bypass donor scepticism. This is not accidental. Decades of behavioural-economics research demonstrate that donor decisions involving identifiable child victims are made with substantially less critical analysis than equivalent decisions involving adults, statistics, or systemic causes.

Reputable anti-trafficking organisations know this. The best of them deliberately publish more methodology, more independent evaluation, and more conservative impact claims than donors emotionally demand — because they understand the field has been damaged by overclaiming and want to operate at a higher standard.

Asking hard questions of a charity working on child sex trafficking does not mean you do not care about child sex trafficking. It means the opposite. Charities that respond well to scrutiny earn donor trust. Those that do not have not earned it.

§ Better-evaluated alternatives

Where donors who care about this issue can give.

The following organisations operate in adjacent anti-trafficking work and have, at the time of writing, published more substantial independent evaluation, methodology, or peer-reviewed research than appears publicly available for Marici. This is not an endorsement; each donor should perform their own evaluation. It is intended to demonstrate what the publicly available evidence base ordinarily looks like in this sector.

  • International Justice Mission (IJM) Works with law-enforcement in 13 countries; publishes annual reports, audited financials, and academic-collaboration studies.
  • Thorn Technology-focused; publishes engineering case studies; specific named platforms (e.g. Spotlight) used by named law-enforcement.
  • Polaris Project Runs the US National Human Trafficking Hotline; publishes methodology and data reports.
  • Apne Aap Women Worldwide India-focused; Ruchira Gupta is a named public figure; documented decades-long operation in Kamathipura.

This site has no relationship of any kind with the organisations listed above, financial or otherwise.