When Greg Squire's team at US Department of Homeland Security Investigations reached out to Facebook for help identifying a 12-year-old girl suffering ongoing sexual abuse, the tech giant declined. The company claimed it 'did not have the tools' to search uploaded family photos for matches, despite possessing sophisticated facial recognition technology that was, at the time, actively deployed across its platform.
Instead, Squire spent months analysing electrical outlets, bedding patterns, and furniture visible in abuse images circulating on the dark web. The case would ultimately be cracked not through Silicon Valley's vaunted technological capabilities, but through the expertise of a brick salesman in America's southwest who could identify a specific variety called 'Flaming Alamo' from a wall glimpsed in the background of abuse material.
The six-year investigation raises uncomfortable questions about what tech platforms mean when they profess commitment to child safety. Facebook's response to the BBC—that it must 'protect user privacy' and 'follow appropriate legal process'—sits uneasily alongside the reality of a child enduring years of rape whilst the technology that might have shortened her ordeal sat unused.
The privacy paradox
Facebook's facial recognition capabilities were hardly secret during this period. The platform routinely used the technology to suggest photo tags and identify users across billions of images. What the company appears to have lacked was not technical capability but willingness to deploy those tools for law enforcement purposes without what it deemed 'appropriate legal process.'
Whether Homeland Security sought proper warrants, or what specific legal barriers Facebook cited, remains unclear. The company has not elaborated on what would have constituted appropriate process in Lucy's case. This opacity is itself revealing. Tech platforms frequently position privacy protections and child safety as competing imperatives, yet refuse to specify what legal standards they require before cooperating.
The practical consequence of this stance falls on investigators like Squire, whose team resorted to Google searches about brickwork. They contacted the Brick Industry Association, whose representative agreed to circulate the image to experts nationwide. John Harp, a brick salesman since 1981, recognised the distinctive pink-cast brick with charcoal overlay immediately.
Manual investigation in the algorithm age
Harp's insight—that the 'Flaming Alamo' variety had been manufactured from the late 1960s through the mid-1980s, and that heavy bricks 'don't go very far' from their production site—proved more useful than any algorithmic assistance. The team had already narrowed Lucy's location to North America based on visible electrical outlets, and had identified a sofa sold only regionally to about 40,000 customers across 29 states.
Cross-referencing the sofa customer list with addresses within 100 miles of Harp's brick factory reduced potential locations to 40 or 50 properties. From there, the team manually searched social media, eventually finding Lucy's photograph alongside an adult woman who appeared to be a relative. They traced that woman's address history and everyone she'd lived with, generating a shortlist of possible homes.
Even then, they couldn't simply knock on doors without risking that the perpetrator would be warned. Instead, Squire's team sent photographs of houses to Harp, asking him to assess whether their construction style and era suggested they might contain Flaming Alamo interiors beneath external cladding. When Harp identified one property as a likely match—also on the sofa customer list—state records confirmed that Lucy lived there with her mother and the mother's boyfriend, a convicted sex offender.
Local Homeland Security agents arrested him within hours. He had been raping Lucy for six years. His sentence exceeded 70 years.
The investigator's burden
What's striking about this methodology is not its ingenuity but its necessity. Elite units tasked with dismantling what the BBC documentary describes as 'the biggest child-abuse forums on the dark web' depend on brick experts and furniture salespeople rather than systematic cooperation from platforms where perpetrators often source or share material.
The human cost of this approach extends beyond the victims. Squire's dedication to Lucy's case—she was roughly his daughter's age, and new images appeared constantly—eventually contributed to alcohol dependency, marital breakdown, and suicidal ideation. His colleague Pete Manning noticed Squire was struggling and encouraged him to seek help. 'It's hard when the thing that brings you so much energy and drive is also the thing that's slowly destroying you,' Manning observed.
Squire now describes exposing his vulnerabilities as essential to continuing work he considers meaningful. The systemic nature of investigator burnout in these specialist units suggests that support mechanisms remain inadequate even as the volume of online child abuse material continues to grow.
Last summer, Squire met Lucy for the first time. She's in her twenties, with sufficient stability to discuss her abuse—something she couldn't manage even a few years ago. She told him that when Homeland Security intervened, she had been 'praying actively for it to end'. Squire wished there had been some way to communicate that help was coming.
Whether tech platforms will reconsider their interpretation of privacy protections versus their stated child safety commitments remains an open question. For investigators tracking abuse material, the current answer appears settled: expect bricks, not algorithms, to provide your breakthroughs.