Project Arachnid is a Canadian pioneering program to take down child sexual abuse images, and this piece reflects on how the project has evolved since 2016, including their work with organisations from six other countries to report abuse.

It also explores and debunks a pervading idea — that tech can save children from, or stop entirely, child sexual abuse. The role of the human eye is crucial to Arachnid's program.

**Key takeaways:
**

  • Arachnid tracks the web and sends takedown notices if abusive images are found. It's now expanded to make sure that images are stored and tracked, to check for patterns between the images, to easier arrest abusers.

  • The tech alone is not enough. Each image Arachnid finds has to be cross-checked by three different analysts, and it's a difficult but necessary job.

  • Arachnid is branching out again to a new program, Shield, for private companies. It's a plug-in, which organisations can use to detect abusive material on their service rather than wait for a takedown notice.

In 2018, tech companies reported over 45 million online photos and videos of children being sexually abused, according to a New York Times investigation which called the problem an “epidemic”.

Project Arachnid, an initiative from the Canadian Centre for Child Protection (CCCP), wants to be part of the cure. The computer program combs the web for explicit images of child sexual abuse.

There are over 11.5 million images on its database, with 100,000 added each month. Beyond the act of abuse itself, many victims struggle to find closure while evidence of their abuse continues to circulate.

Arachnid is trying to stop the flow of online material. Its work has changed and evolved, including bringing in the private sector and collaborating with countries beyond Canada. But analysts work to ensure that images are what the code suggests. Can tech ever be a replacement for the human eye?

Arachnid’s origins

Arachnid began as a traditional tipline, where citizens could use an online form to report images. It depended primarily on reports from Canadian citizens and law enforcement.

“Every report felt like a failure,” said Lloyd Richardson, the Centre’s Director and a driving force behind Arachnid’s development, speaking to Apolitical in 2017. ‘When we get reports from members of the public, it means these images already have some degree of traction.”

So they built Arachnid version 2.0, a program tracking the web and sending takedown notices if abusive images were found. If technology helps the proliferation of paedophile networks, they reasoned, then technology can become part of the solution.

“A lightbulb went off in our heads when we realised its potential"

Arachnid works using hash values: strings of code produced by running images through scanning software —a kind of digital fingerprint.

The program scans online images to produce hashes before checking them against a master list of known sexual abuse images.

The master list is “triple-vetted”—seen by three human analysts who confirm that an image depicts illegal sexual contact with a minor.

To avoid “grey areas”, Arachnid deals only with “the worst of the worst”, Richardson explained. It avoids images depicting people of ambiguous age, or those generated consensually by under-18s.

The next iteration

In 2019 the project is entering a new phase, one which could profoundly strengthen the ability of government and tech companies to combat child abuse imagery long term.

“A lightbulb went off in our heads,” said Richardson, “when we realised its potential”.

Arachnid still crawls the web, but since 2018 the program has complemented this with more targeted scraping. This stores and processes the data from the images, to see if it has a similar hash value to others in the database.

“Originally, we stored the images temporarily, until they were removed,” said Richardson. “Now, we’re storing them long term, to establish connections with the new images that are coming in — so they can be removed quicker.”

“It’s graphic, disturbing, heinous, and infuriating. But you take that anger, that rage, and all the other emotions you are feeling, and use it to fuel your drive.”

It then spots potential repeat offenders, making them easier to prosecute, and letting law enforcement step in quicker — and with more evidence.

Despite its innovations, Arachnid still today relies on its team of human analysts to vet its backlog of images. Of the 91 billion images that have now been processed by November 2019, over 13 million images have been triggered for analyst review.

“Megan” in 2017 was a senior analyst examining images referred by Arachnid. She spoke to Apolitical under the condition that we omit her surname.

“It’s graphic, disturbing, heinous, and infuriating,” she said. “But you take that anger, that rage, and all the other emotions you are feeling, and use it to fuel your drive.”

“We’re not going to solve this problem by throwing an algorithm at it"

Arachnid protects its staff, but it’s still a tough job. “People take breaks, there are shifts, and we have counsellors on hand if anything is particularly traumatic,” Richardson explained.

The analysts’ work disputes the narrative that tech can be a shortcut to success for supporting child abuse victims, above more traditional preventative methods like parenting programs.

“We’re not going to solve this problem by throwing an algorithm at it,” Richardson said, emphasising the value of human analyst interaction within the service. “If it were an easy problem to solve, we’d throw our hands up in the air and solve it tomorrow.”

Going global

Arachnid continues to work with its more traditional tipline alongside its other work, and is now in the early phase of bringing other international tiplines on board. Six nations are already signed up, including Colombia and Croatia.

Because different services cannot share the illegal imagery between them, it’s harder to collaborate. But analysts have now built a platform to help tiplines communicate — Arachnid Orb. Orb reduces the duplication in assessment time and increase the number of notices that can be sent through Project Arachnid.

But it takes time to get other tiplines involved in the project. A complex vetting process is in place to ensure tiplines are the right fit for the platform.

Private sector collaboration

The cross-border nature of online child protection law often requires enforcement agencies to reach out beyond their domestic jurisdictions. But of the 40 countries in the Economist Intelligence Unit’s Out of the Shadows Index, only eight have a dedicated budget to counter child sexual exploitation.

To build better links with industry, Arachnid has developed its “Shield” program. Companies can plug in Shield’s free software to detect abusive material on their service rather than wait for a takedown notice. This reports the URL to Arachnid automatically and speeds up its removal.

“It’s depressing in some ways. But we now have a better idea of the scope of the problem. I can’t emphasise enough how large it is.”

But even when laws are in place, and government and tech companies are working side by side, Richardson reflected that a continued problem persists: children themselves are scared to report online child sexual exploitation over fears that more people will look and find evidence of their abuse.

Faced with so many competing challenges, running a project like Arachnid can sometimes feel insurmountable. And as Arachnid grows, so does its body of images. “It’s depressing in some ways,” said Richardson, reflecting on the program’s progress. “But we now have a better idea of the scope of the problem. I can’t emphasise enough how large it is.”

Indeed, the internet’s biggest strength is also its biggest weakness when it comes to stopping the spread of these images: it’s a global resource.

So rather than other governments trying to replicate Arachnid’s approach, Richardson suggests collaborating. “Rewriting Arachnid’s program is counterproductive,” he said. “Fund your own internal tiplines, encourage local companies to use our Shield program, and get in contact.” — Emma Sisk and Ed Siddons

Test what you learned

Try and answer without looking back at the article: