70 Million Warnings

Proof We Can Stop This Before It Starts

---

Editor’s note, added [15 May 2026]:

Since publishing this piece, I was contacted by Sarah Smith, author of the Project Intercept impact report and lead on the project itself. She raised an important and fair correction about how I framed the 1% click-through rate, and she is right.

I described it as a design problem. That was too narrow.

Project Intercept’s warning messages serve two functions: they deter, and they signpost. I only measured the second one. The deterrence function is harder to quantify, but the data exists. An evaluation of Project Intercept’s chatbot intervention on Pornhub UK found that 82% of users desisted from searching after receiving a single warning message. That is not a footnote. That is the primary mechanism doing exactly what it was designed to do.

Not everyone who stops searching needs to click through for help. Some people need the behavioural nudge, and that is enough. Click-through matters for those who are ready to seek support, but it is the wrong headline measure for the project as a whole.

I have updated the relevant section below to reflect this. The correction is mine. The work Sarah and the Lucy Faithfull Foundation are doing remains, as I said, genuinely remarkable.

---

I want to start this one with something I don’t experience often enough.

I want to start with hope.

That doesn’t come easily in this work. After eight years investigating digital crimes involving children, years that cost me more than I knew at the time, I’ve got a fairly low threshold for optimism when it comes to what’s happening online. But every now and then, something happens that makes me think, right, this is proof that the fight is working.

This is one of those things.

But I’m also not going to let anyone off the hook, because buried inside the hopeful headline is a number that demands a conversation.

What Actually Happened

In the past two years, more than 70 million warning messages were sent to people who were searching online for illegal images of children.¹

Seventy million. Let that number breathe for a second.

These warnings are part of something called Project Intercept, run by the Lucy Faithfull Foundation, a UK charity that has been doing critical child protection work for decades. The way it works is straightforward in principle, and genuinely impressive in practice. When someone uses search terms online that suggest they’re looking for illegal content involving children, or clicks on a link previously flagged as containing such material, they receive a message, not a block, not an arrest. A message that says what they’re doing is illegal, explains the harm, and offers them a route towards help.

The tech partnership behind this includes Meta, TikTok, Google, and platforms spanning gaming, streaming, dating and AI. Twenty-two different warning messages, reaching users in 131 countries. At its peak, more than 95,000 alerts were triggered every single day.²

That is an extraordinary thing, I want you to know it exists.

The AI Factor

I’ve spent a lot of time talking to parents, teachers, and anyone who’ll sit still long enough, about what AI has done and continues to do to the threat landscape for children.

Not because I want to scare people. Because understanding what we’re dealing with is the only way to fight it properly.

The Lucy Faithfull Foundation’s chief executive said it best: “The need has never been more urgent, particularly as new AI technologies accelerate the spread of online child sexual abuse.”

I’ve been saying a version of this since I started blogging. The technology that used to make the creation of illegal imagery difficult, requiring real-world access to victims, has been fundamentally disrupted. AI has lowered the barrier so dramatically that the scale of what’s being produced and circulated is growing in ways that are genuinely hard to track, let alone address.

And here’s what I find both infuriating and galvanising in equal measure.

The same AI capabilities being used to create and spread harm are also being deployed to stop it. Project Intercept is living proof of that. Intelligent systems, running quietly in the background on the platforms your children use every day, identifying the patterns, catching the searches, stepping in before harm can progress.

This is what prevention at scale looks like. And I want more of it.

The Number I Got Wrong

I want to be straight with you, because I always try to be, that means owning something.

When I published this piece, I framed the 1% click-through rate as a problem, possibly even a design failure. Sarah Smith, the author of the Project Intercept impact report and lead on the project itself, reached out to correct me. She was right to do it.

Here is what I missed.

Project Intercept’s warning messages do two things: they deter, and they signpost. I only measured the second one.

The deterrence function is the primary mechanism. When someone receives a warning message, the majority stop searching. Full stop. A chatbot evaluation on Pornhub UK found that 82% of users desisted after a single warning message.

That is not a footnote to the story. That is the story.

Not everyone who stops needs to click through to access support. Some people need the behavioural nudge, and nothing else. The click-through rate only captures the people who are both concerned about their own thoughts or behaviour and ready to seek help in that moment. That will always be a small proportion, and it is supposed to be.

Professor Sonia Livingstone from the London School of Economics noted that the click-through figure seems low given the scale of the warnings sent.3 That observation has merit in isolation, but Sarah’s point is the important one: measuring success only by click-through is to miss the bigger picture. The invisible deterrence effect is still working. It is just harder to count.

So to be clear: 700,000 people being directed towards support is real and meaningful. And 70 million warnings disrupting behaviour in the moment, with the majority of people desisting on contact, is the intervention doing exactly what it was designed to do.

I was too quick to call that insufficient. The correction is mine.

⚡Please don’t forget to react & restack if you appreciate my work. More engagement means more people might see it. ⚡

What This Means for Our Children

Here is what I want you to actually take from this, beyond the statistics.

This story is proof that technology can be deployed proactively, before harm happens, rather than just reactively after the damage is done. We spend a lot of time, rightly, holding platforms accountable when they fail. Project Intercept is what it looks like when they commit.

There is a version of the internet where the architecture itself pushes back. Where a journey towards harm is intercepted before it reaches a child. We are not there yet. But we are moving in the right direction.

It also confirms something I’ve believed since this blog began, the people who pose a risk to children online are not a fringe problem happening in places the rest of you will never see. They are on the same platforms your children use. They are triggering 95,000 warnings a day. That number is uncomfortable, and it is meant to be.

A Word on Accountability

The Lucy Faithfull Foundation has called on more tech companies to join and scale what works. I’m adding my voice to that.

Google, Meta, TikTok: you’re on the right side of this particular line. The partnership with Project Intercept represents a genuine commitment, and the results, imperfect as they are, represent real harm reduction.

But 1% is not a finish line. The click-through rate tells you the messaging isn’t reaching the people who most need to change their behaviour. The infrastructure exists. The intervention points exist. The follow-through needs work, and you have the capability to improve it.

Every platform not on that partnership list, under the Online Safety Act, you have obligations. Ofcom has powers it has not yet fully exercised. The question is whether it does.³

What You Can Do Right Now

I’m not going to pretend this is light reading. It isn’t. I know that.

But I want to leave you with something concrete.

1. Keep the conversation going at home. A child who feels they can talk to a trusted adult about anything they’ve seen or been asked online is safer. It sounds simple. It is, in my view, the single most protective thing you can do. Not monitoring software. Not content filters. Open, honest, non-judgmental conversation.

2. Know the resources exist. The Lucy Faithfull Foundation’s Stop It Now service (stopitnow.org.uk) is for anyone concerned about their own thoughts or behaviour, or about someone else’s. It is confidential. It exists because early intervention works.

3. Report what concerns you. If something on a platform worries you, report it through the platform’s own tools. If the response is inadequate, escalate to Ofcom. These systems only work when people use them.

4. Talk to your children about AI. Not to frighten them. To arm them. People online are not always who they claim to be, and AI has made that harder to spot than ever. That conversation needs to be happening now, if it isn’t already.

If any child is worried about something they’ve seen or been asked online, they don’t need to face it alone.

As always, thank you for your support. Please share this across your social media, and if you do have any comments, questions, or concerns, then feel free to reach out to me via the Social page, as I am always happy to spend some time helping to protect children online.

Remember that becoming a paid subscriber means supporting a charity very close to my heart and helping it do amazing things for people. Childline, I will donate 100% of paid subscriptions collected every six months, as I don’t do any of this for financial gain.

Want to Support Great Causes? (And Get Absolutely Nothing Else?)

If you or a child you know needs support:

Childline: 0800 1111 | childline.org.uk

Available 24/7, 365 days a year. Free, confidential, and here for every child.

---

1. Sky News, “More than 70 million warnings sent to people searching for child sexual abuse content,” 13 May 2026. https://news.sky.com/story/more-than-70-million-warnings-sent-to-people-searching-for-child-sexual-abuse-content-13543484 (accessed 13 May 2026). Note: verified via multiple secondary sources

2. Lucy Faithfull Foundation, Project Intercept data release, May 2026, as reported by BBC/Sky News/AOL. Figures cover 2024-2025. https://www.lucyfaithfull.org.uk (accessed May 2026).

3. UK Online Safety Act 2023. Ofcom enforcement framework for illegal content duties. https://www.ofcom.org.uk