This AI can spot signs of human trafficking in online sex ads

Ex-military and intelligence pros are using computer vision to scan illicit sex ads and bust the pimps who enslave innocent women and girls.

This AI can spot signs of human trafficking in online sex ads
[Images: Oleg Polyashenko/iStock (frames); neyro2008/iStock (grid); Raphael Lovaski/Unsplash; henri meilhac/Unsplash; Ioana Cristiana/Unsplash; Rafael Saes/Unsplash; Jana Sabeth/Unsplash]

Molly is a “new bunny” in Atlanta and loves her job selling sex, domination, and striptease shows to men. So says an ad on one of the popular websites listing “escorts” or “dates” in the U.S. But chances are Molly has been forced into this work. She’s likely a victim of sex trafficking—kidnapped, brainwashed, or blackmailed into prostitution. The proof is right in the ad, if you know what to look for.


“The sad truth is that it’s happening in every city in America,” says Kara Smith, a senior targeting analyst with DeliverFund, a group of former CIA, NSA, special forces, and law enforcement officers who collaborate with law enforcement to bust sex trafficking operations in the U.S. The online ad for Molly provides clues that she’s performing against her will and is not an independent sex worker who keeps all the money she earns. For instance, she’s depicted in degrading positions, like hunched over on a bed with her rear end facing the camera.

“Any self-respecting woman, even if she is a prostitute, she’s going to sell herself as a hot commodity,” says Smith. “She’s going to sell herself with a little bit more class, and she’s going to be very picky” about the kind of clients she attracts.

Bruises and bite marks are other telltale signs for some victims. So are tattoos that brand the women as the property of traffickers—crowns are popular images, as pimps often refer to themselves as “kings.” Photos with money being flashed around are other hallmarks of pimp showmanship.

Seasoned experts like Smith can spot these markers, but they no longer have to do it all manually. About a year ago, DeliverFund got an offer from a computer vision startup called XIX to automate the process. The company’s software scrapes images from sites used by sex traffickers and labels objects in images so experts like Smith can quickly search for and review suspect ads. Each sex ad contains an average of three photos, and XIX can scrape and analyze about 4,000 ads per minute, which is about the rate that new ones are posted online.

“What we would have to do without XIX’s abilities would be to just go to the illicit website and scroll for hours and hours,” says Smith. “Their tool is very important because it gives us the ability to sift through millions of data points while looking for information more quickly.”


Technology meets problem

The relationship between the organizations began in 2018 when XIX founder and CEO Emil Mikhailov saw a Vice mini-documentary about DeliverFund’s executive director, Nic McKinley. After working as an Air Force special ops rescuer for 11 years, McKinley became a CIA operative, where he encountered child trafficking cases. In 2014, he founded DeliverFund, a Dallas-based nonprofit that now has over 30 people spread across the U.S.

Working with just three people, and no image-rec tech, for its first three years, DeliverFund was still able to produce intelligence that led to the arrest of four pimps. The organization began staffing up in 2018, when it helped law enforcement bust 19 traffickers and free 17 victims. In the first nine months of 2019, DeliverFund contributed to the arrests of 25 traffickers and 64 purchasers of underage sex. Over 50 victims were rescued in the process. That increased impact is due in part to the group’s use of technology.

DeliverFund also provided intel to assist the takedown of, which had become the top place to advertise sex for hire—both by willing sex workers and by pimps trafficking victims. In a controversial move, was seized and shut down in April 2018 by a joint government operation including the FBI, the Justice Department’s section on child exploitation, and several state attorneys general. (As a coup de grâce, DeliverFund moved its headquarters into the former Backpage office in Dallas.)

After watching the Vice documentary, Mikhailov contacted DeliverFund and initiated a conversation about using computer vision to help scan photos. “When Nic was describing their approach, I figured that we may be able to apply our expertise to this problem,” he explains. “I just cold emailed him saying, ‘Hey, we can [label] the content of the images. Will that be helpful?'”

Pro-bono work for DeliverFund was the perfect first gig for XIX, says Mikhailov. The company emerged from the winter 2017 cadre of startups backed by the Y Combinator accelerator program in Silicon Valley. It originally made Android software that predicted what users want to do—like offering to order an Uber when they leave work.


In 2018, XIX pivoted to computer vision but struggled to focus it into a business. “Engineers, especially researchers, tend to work on fun stuff that is rarely being applied in the real world,” says Mikhailov. “The work with DeliverFund shaped the requirements of what is necessary to build something that people actually want.”

XIX has since picked up a handful of paying clients. One, for instance, uses the company’s vision tech to power an NSFW content blocker for offices. Another is a media company that uses the tech to search and retrieve images and videos from its archive.

Data in action

XIX’s computer vision is a key tool in a digital workflow that DeliverFund uses to research abuse cases and compile what it calls “intelligence reports” for law enforcement to act upon. Smith reckons that digitizing the entire process cuts the time to develop an intelligence report from 22 days to just six hours. (DeliverFund has partnerships with several other tech companies whose services help automate its work, including IBM and LexisNexis.)

XIX’s software scrapes images and other data from ads, creating a database that DeliverFund can search for an evolving list of criteria.[Screenshot: XIX]
“We just don’t go out there and start developing out cases unless we know we have law enforcement there to take the intelligence that we provide,” says Smith. “They are the door kickers. They have the badge, and they have the ability to arrest those traffickers and save those victims,” she says.

[Screenshot: XIX]

DeliverFund has provided intel to 63 different agencies across the U.S. Some are local, such as the Houston Police Department. The organization also has relationships with the attorney general’s offices of Montana, New Mexico, and Texas, allowing it to coordinate with law enforcement statewide. DeliverFund has also worked with the Department of Homeland Security in several states.


DeliverFund provides free training to law officers on how to recognize and research abuse cases and use its digital tools. Participating agencies can research cases on their own and collaborate with other agencies, using a DeliverFund system called PATH (Platform for the Analysis and Targeting of Human Traffickers).

Targeting the source

DeliverFund is adept at building partnerships with government agencies because its staff has extensive government experience. Smith, now based in Albuquerque, spent six years as an Air Force intelligence analyst, deploying to both Afghanistan and Iraq. She later worked for about three years each for the National Security Agency and the FBI on counterterrorism projects. That experience informs the way she approaches her current job, which she started about two years ago, after McKinley recruited her on LinkedIn.

“When it comes to counterterrorism, we go after what would be the most lucrative target, the one that would be the biggest threat to us,” she says. That’s not the individuals planting bombs, but the people who finance their work. “Now with counter human trafficking, we look at taking out the human trafficker, because without a human trafficker, there wouldn’t be any victims being trafficked.” Each pimp has an average of three to five victims, known as a “stable,” she says.

XIX’s tool helps DeliverFund identify not only the victims of trafficking but also the traffickers. The online ads often feature personally identifiable information about the pimps themselves.

A pimp spends 6 to 12 months “grooming” each victim into the trade, says Smith. Pimps may lure women and girls with gifts and promises of a lavish lifestyle, then blackmail them after shooting compromising photos or videos. Some pimps encourage drug use (typically heroin) and withhold more unless the victim brings in a certain amount of money each day. About half of trafficking victims worldwide are minors, according to the Human Trafficking Institute.


Despite a common perception that sex trafficking victims are foreigners, the overwhelming majority are U.S. citizens, says Smith. “In the hundreds of cases I’ve worked, I’ve only seen three cases where there’s actually been victims that were foreign,” she says. “It is happening here to our own citizens, and it’s being done by our own citizens.”

Human traffic is a gruesome business, and combating it is a daunting task, especially given how rapidly it can spread online. But Smith is animated with a spirit of optimism, based on the accomplishments she and her teammates have made—accomplishments heavily aided by technology.

She uses words like “awesome” to describe fruitful collaborations with law enforcement that help save victims. “It’s our goal to assist them in any way that we can, at no cost, to help develop them and build them up and go after the problem,” she says.

Contact me with tips about human trafficking or any other topic at or encrypted at, or reach out on Twitter or Facebook.

About the author

Sean Captain is a Bay Area technology, science, and policy journalist. Follow him on Twitter @seancaptain.