Whistleblower: Compassion Ignored Years of Warnings About Child Photos and AI Sextortion

A whistleblower disclosed that Compassion International continued posting children's photographs online despite internal warnings that artificial intelligence tools enable predators to manipulate the images for sextortion. Andrew Vezey, a former Compassion technology specialist, first alerted leadership to AI photo manipulation risks in 2023; in early October 2024 the organisation's safeguarding team confirmed a Compassion child had been victimised through AI-based sextortion. Separately, in 2024 Vezey helped develop an AI content-screening program that detected 25,000 suspected grooming cases in letters from Compassion sponsors to children and young adults, yet the organisation maintained its policy of publishing children's faces online, citing donor relationship requirements.

AI-generated and deepfake sexual abuse material now circulates at scale, and conventional content moderation cannot distinguish synthetic imagery from records of contact abuse until after publication and distribution. Detection systems built to recognise coercion, escalation and age-inappropriate contact patterns in live conversation can flag a grooming attempt or sextortion demand as it unfolds, not after an image has been weaponised. Guardii anti-sextortion and anti-grooming filters, developed by a Meta Business Partner and backed by Startmate, analyse the relational dynamics in a direct-message thread in real time, surfacing the child at risk to a parent, school or agency before explicit material is created or shared, and blocking further contact from the offender without requiring the child to recognise or report the manipulation themselves.