An AI to trace bitcoin transactions to help detect sexual trafficking.

A few weeks ago, New Scientist shared an excerpt of ‘I am Jane Doe’, a documentary about the story of many young women who have fallen into prostitution after being trapped by newspaper ads. Said like that, we seem far from the title of the article, but promised, we will surely arrive at the doorsteps of bitcoin.

The extract taken by New Scientist retraced the (real) case of the testimony of Kubiiki Pride, a mother whose 13-year-old daughter had disappeared. She found her traces through photos of her daughter published on the website called,, a website that hosts hundreds of pages of advertisements dedicated to the sex industry, majority of which are illegal.

Similar to Craiglist, Backpage housed more than 70% of the US online sexual advertising market in 2012. It took 270 days for Kubiiki Pride to find her daughter, who had been drugged and beaten by the sexual trafficker who seized her.

For the authorities, it is very difficult to cope with this kind of case because a lot of ads are posted every day, making each girl a drop of water that is almost impossible to find. For this reason, researchers at the University of California at Berkley have attempted to take the problem differently, to try a new approach using technological means, Artificial Intelligence, Machine Learning, and cryptocurrency, to be specific.

One of the researchers, Rebecca Portnoff, said: “The Internet has facilitated a lot of methods that traffickers can take advantage of.” They can easily reach a large audience and generate a lot of content without having to reveal themselves. In this case, use the same weapons as the traffickers to nail them in the bud.

Since 2015, every advertising transaction made on is done in bitcoin and is registered in a manner accessible to all. Although identities are never revealed, transactions are associated with a portfolio. This is where the team of researchers comes in: the University has developed artificial intelligence that uses machine learning to track money. The purpose of the AI is to search the various pages of ads to identify the portfolio of each ad, and to detect particular or repeated transactions.

Also, ads with a similar editorial style are subject to tool analysis. If transactions and announcements appear abnormal, they are forwarded to the authorities for scrutiny. The real advantage of this tool is the automation of a task that requires an enormous amount of time and resources for the authorities, where AI can optimize research.

The research team presented the AI at a Canadian data mining conference after a 4-week trial period. Of the 10,000 ads, the tool correctly identified approximately 90% of ads with the same author, with a false identification of only 1%. One of the portfolios analyzed was an author with $ 150,000 of sexual advertising, a possible proof of sexual trafficking. If the results are successful, the tool can be extended to the rest of the country.

For the trafficker who abducted Kubiiki Pride’s daughter, he was eventually arrested and given a five years prison sentence. With this new tool, prosecutions against sexual traffickers could increase.

Source: New Scientist.