Skip to content
Thorn

2025 Impact

Unstoppable progress in child safety

IMPACT AT A GLANCE

5,303,413

Suspected CSAM files found

1,369,024

Instances of suspected child exploitation found

415.3 billion

Files processed by our customers

86

Number of tech companies using Thorn’s CSAM detection products

15,038

Research report downloads

1,046

Agencies using Thorn’s child victim detection tools

Letter from the CEO

In the pursuit of child safety, there is no substitute for bold action. Meaningful progress requires all of us, working together, to build the online safety systems that should have existed from the very beginning. That system is a multilayered, ever-evolving digital safety net.

And we are building it every day.

I’m proud to say that in 2025, we significantly scaled our impact across four key pillars of child safety. Together, these pillars represent how we drive meaningful change and protect children at scale.

Through research and insights, we map digital threats and share critical findings that strengthen the broader child protection ecosystem. Last year, we released four key research reports that expanded our understanding of deepfake nudes, sextortion and how youth experience risk and harm online.

Through technical innovation, our world-class machine learning engineers harness the power of AI to improve detection tools and safety technology. In 2025, we introduced new capabilities, like online grooming detection, that provide a more robust toolkit for both technology platforms and investigators.

Through our child victim identification pillar, we develop digital forensics tools that help investigators find children who are being sexually abused more quickly. The rollout of Thorn Detect in 2025 marked an important step forward, providing investigators with new technology in the race against time to identify and protect victims.

Through our platform safety efforts, we equip technology companies with purpose-built solutions and expert guidance to detect and address child sexual abuse and exploitation. This year, more than 86 companies, 43% more than 2024, utilized Safer for proactive detection.

The digital world wasn’t built with child safety in mind, but that doesn’t mean it has to stay that way. Every step forward — in awareness, in understanding, and in technology — has the potential to alter the trajectory of a child’s life.

When I reflect on the numbers in this report, I see accelerated impact at a scale we only dreamed of a decade ago.

In 2025, our technology helped uncover more than 5 million files of suspected child sexual abuse material across platforms, while also surfacing more than 1.3 million lines of text associated with suspected child exploitation, including signs of grooming, giving platforms the potential to identify and interrupt harm earlier.

At the same time, new technology like Thorn Detect is helping investigators in 40 countries quickly detect suspected child sexual abuse images and videos, ultimately helping them identify children faster and remove them from harm.

Through our work advancing Safety by Design for generative AI, many companies are committing to building technologies with safeguards that reduce the risk of misuse from the start.

Together, these advances aim to shift the timeline of intervention — from limiting the circulation of harmful content to detecting risk earlier and responding faster when children need it most.

These numbers represent real moments when children — kids in our families and communities — are at risk, when earlier visibility can mean the difference between harm continuing and harm being stopped.

Protecting children is everyone’s responsibility. And the progress we’ve made this year is only possible because of a community of supporters who believe, as we do, that technology can and must be part of the solution. Your support fuels the research that helps us understand emerging threats. It powers the innovation that enables faster detection. And it strengthens the tools that help investigators find children and bring them to safety.

As we look ahead, we know the challenges will continue to evolve. But so will we. Together, we are building a digital safety net that can transform how children are protected for generations to come.

Thank you for being part of this work.


Julie Cordua, CEO

Be a force for good.

Your donation helps protect children every day. Through research and technology, we can transform how children are protected in the digital age.

Donate Today

Research & Insights

Technology never stands still. It seems that nearly every day, a new app, platform, or social trend intertwines with the lives of young people. To keep up with the ever-evolving digital world — and the harms that stem from it — we must understand the lived experiences of young people. Research is an essential tool for achieving this.

For nearly a decade, Thorn has conducted youth-centered research to understand how young people are experiencing harm online. This work helps identify emerging trends in online life and advance the child safety ecosystem.

Four keystone research reports in 2025

Deepfake Nudes and Young People: Exploring young people’s awareness of deepfake nudes, experiences with them, and their involvement with creating such content.

Commodified Sexual Interactions Involving Minors: Examining how technology has changed the concepts of “gifts” and “payment” as it relates to sexual transactions and exploitation.

Sexual Extortion and Young People: Taking a look at sextortion dynamics involving young people to see how often and in what forms this abuse takes place.

Youth Perspectives on Online Safety, 2024: Thorn’s annual report of Youth Perspectives represents the sixth year of data collection on youth behaviors and attitudes toward online safety.

15,038

Research Report Downloads

2,248

Youth and Young People Surveyed

Informing the world with research and insights

Transforming data into insights helps raise awareness across the child safety community and the general public. We pursue media opportunities to share our findings in ways that make them impactful and actionable. To help mitigate potential harm, it’s essential to raise awareness and understanding of it.

Family fights to end “sextortion” scams after son’s death.
CBS News

How AI is being used to create explicit deepfake images that harm children.
PBS NewsHour

Teens cope while adults debate harms of fake nudes.
Ars Technica

1,124,000,000

Research-Related Media Impressions

Safety by Design for generative AI

In early 2024, Thorn collaborated with All Tech is Human to launch our Safety by Design initiative for generative AI. We brought some of the most influential AI companies to the table and secured commitments to prevent the creation and spread of AI-generated CSAM. In 2025, Safety by Design continued to gain traction, both within the tech industry and with standard-setting organizations.

One Year of Safety by Design. In 12 short months, this initiative yielded impactful results, with 11 companies committed to Safety by Design principles. Their self-reporting included:

  • Companies detected and blocked hundreds of thousands of attempts to generate harmful content.
  • Hundreds of models that were built to create child abuse imagery were removed from platform access.
  • Multiple training datasets at leading AI companies were filtered for CSAM and other abusive material.
  • Millions of prevention and deterrence messages were surfaced in response to violative prompts.
  • Companies invested significant resources in new protective technologies and research.

 

Standards

UK AI Security Institute (AISI). Thorn co-authored a safety protocol with the UK AISI to create a framework to prevent AI-generated CSAM. This represents a significant step forward in aligning Thorn’s expertise with state-backed safety organizations.

The Institute of Electrical and Electronics Engineers (IEEE). Thorn led an IEEE working group to draft the first international standard embedding child protection across the AI lifecycle. In September 2025, the working group advanced the standard to be voted on. Embedding Safety by Design as an IEEE standard will provide a solid foundation for safer development, deployment, and maintenance of generative AI models.

National Institute of Standards and Technology (NIST). As a part of the U.S. Department of Commerce, the NIST’s standards are highly respected and globally influential. NIST’s AI 100-4 standard is focused on reducing risks from synthetic content, including AI-generated CSAM. Thorn was invited to provide early feedback and deep guidance on this groundbreaking AI standard that helped shape new U.S. best practices on reducing risks from synthetic content.

Informing policymakers

A key element of informing the child safety ecosystem is keeping policymakers and regulatory bodies up to date on emerging harms and how technology can help protect children. Our expertise helps key decision makers — particularly in the US, UK, and across the EU — make informed decisions about legislation and regulatory guidance that can keep children safe. 

In the US, we engage regularly with congressional champions of our issue space to support important legislation. In 2025, we supported the Take It Down Act (which passed into law in May 2025), as well as the STOP CSAM Act. We also hosted research briefings for U.S. policymakers on emerging harms in our issue space, in addition to research and technical briefings for policymakers in the UK and EU. 

EU AI Act (2025): Thorn was one of the civil society organizations contributing to the EU AI Act Code of Practice, the EU’s first major regulatory tool for AI. Our input helped secure key wins like requiring companies to document how they remove CSAM from training data and treat CSAM as a systemic risk.

88

Policy discussions

5

Technical briefings

12

Speaking engagements

Technical Innovation

The internet wasn’t built with child safety in mind, and predators have taken advantage of that reality. As digital spaces rapidly evolved, critical safeguards lagged behind, creating conditions where child sexual abuse and exploitation could scale at an unprecedented rate. 

But Thorn, our partners, and our supporters see a solution. The very technology that has been used to exploit children can, and must, be harnessed to protect them. 

We’re building a technical foundation to scale child protection through advanced tools and ecosystem-wide collaboration. By contributing to shared efforts across the child safety community, we help accelerate innovation beyond our own products. This integrated approach strengthens both victim identification by law enforcement and platform safety, creating a layered system of protection for children.

New technology launched in 2025

With every new launch and release, we advance the digital safety net. Whether developing AI-powered detection tools, improving existing tools, or scaling deployment across the industry, every step forward protects more children around the world. 

Text-based grooming detection

The technical innovation team developed text-based grooming detection, which was deployed on our purpose-built platform safety solution, Safer Predict. Now, platforms can flag grooming behavior and language that can indicate early-stage sexual exploitation or abuse of a minor.

Spanish-language detection

Safer Predict expanded the communities it can serve with the addition of Spanish-language text detection. This allows platforms to expand their ability to detect suspected text-based child sexual exploitation, like sextortion and child access. This expansion is a perfect example of unlocking existing technology to scale protection for countless additional youth.

New versions of CSAM image and video classifiers

Tools based on machine learning aren’t a “build once and forget it” exercise. Our machine learning engineers are consistently maintaining, refining, and improving our models to enhance performance, increasing their ability to detect new or previously unknown CSAM, which helps investigators identify more children faster.

Collaborations

We work with partners to help advance technical innovation across the child safety ecosystem. These collaborations help apply Thorn’s expertise to different aspects of child safety.

We are proud to support the ongoing work and impact of the INTERPOL DevOps working group in disrupting child sexual exploitation and abuse. Thorn participates in INTERPOL DevOps activities to help develop scalable technology to combat online child sexual exploitation. Alongside many other law enforcement, nonprofit and private-sector partners, we work to build tools that help investigators identify victims faster, reduce exposure to harmful content and improve investigative efficiency.

This work contributes to protecting vulnerable children across the globe and highlights the importance of international collaboration and technical innovation in combatting child abuse in the digital age.

Other collaborations include:

  • Our data science team contributed key insights to a Stanford Center for Research on Foundation Models position paper on AI flaw reporting.
  • We presented at the “AI For Public Good” forum organized by AI Portland.
  • We explored how to mitigate risk in AI with the Emerson Collective.
  • Thorn’s Dr. Rebecca Portnoff presented about child safety for generative AI at several events, including the Google Growing Up in the Digital Age Summit, the AI Conference, and an All Tech Is Human Fireside Chat.

Where continuous innovation meets child protection

Breakthrough technology doesn’t happen in a single version. It’s built through continuous iteration.

In 2025, Thorn’s machine learning team developed over 220 model versions across 15 active projects. Each iteration represents critical progress — testing, refining, and improving how we detect suspected abuse faster, support investigators, and protect children.

That scale of innovation — driven by a focused, experienced team — is what allows Thorn to keep pace with evolving threats and deliver real-world impact.

Child Victim Identification

Child victim identification is a needle-in-a-haystack forensics challenge with children’s lives on the line. For every case, investigators must often review vast quantities of images or videos to find the child in active harm’s way. Searching thousands to hundreds of thousands of files for identification clues is an urgent, overwhelming task. Needless to say, these investigations can take weeks or months. And that is too long when a child is in danger.

Thorn deploys our technical innovation to the task of helping investigators find child victims more quickly. Using state-of-the-art machine learning, these forensic solutions process files faster than humans can manually, turning what used to take days into hours. This helps them to focus their resources on identifying and finding victims and arresting perpetrators.

Thorn Detect

In 2025, we launched Thorn Detect, an innovative tool that represents a major advancement in detection technology, featuring CSAM classifiers that provide investigators with four key advantages:

  1. Quickly detect suspected CSAM. Thorn Detect’s machine learning classification models are a critical component in helping investigators quickly detect suspected abuse images and videos.
  2. Prioritize the most critical files.  Classification models help investigators rapidly triage large volumes of content and prioritize review of the most critical files.
  3. Work more cases and locate more children. With limited resources and investigators available, this technology gives investigators more time to locate child victims and allows them to work more cases.
  4.  Reduce exposure to traumatic content. Thorn Detect helps investigators avoid burnout from the emotional burden of traumatic content.

To help illustrate the urgency of investigating these active cases, we invited Detective Michael Fontenot to share his experience.

1,046

Agencies using child victim detection tools

40

Countries where child victim detection tools are deployed

89%

Of investigators reported that Thorn’s CSAM classifier saved them time in their work

The impact of Thorn’s technology isn’t just about efficiency for investigators. It’s about children’s lives. Thorn Detect gives us back time to work more cases. Time to find more victims. Time to stop the abuse sooner.

Det. Michael Fontenot, Dallas Police Department

Investigating the Dark Web

The dark web is a hidden corner of the internet accessible through specialized browsers and designed to provide anonymity and privacy. It has become a haven for criminal activity, including the abuse and exploitation of children. For investigators, the dark web poses the challenge of tracking down offenders who believe they’re operating beyond the reach of law enforcement. Thorn’s tools represent transformative solutions for investigators working these demanding cases.

This year, former dark web investigator Pete Manning shared what it’s like to investigate these cases.

Platform Safety

Technology companies play a key role in protecting children from abuse and exploitation. Our technology empowers them to combat the spread of child sexual abuse material and detect exploitation. As experts at the intersection of child safety and emerging technology, Thorn is uniquely positioned to lead this critical element of the child safety ecosystem. With purpose-built tools and expert consulting, we can help technology companies turn their platforms into unwelcoming environments for abuse.

Safer: Thorn’s platform safety solution

Safer reflects Thorn’s expertise in child safety, paired with the technical strength of experienced machine learning engineers and developers. This innovative toolset helps platforms identify known CSAM, which is often shared virally, and suspected CSAM, which can be related to active abuse, along with possible text-based threats like grooming and sextortion.

For Safer, 2025 was a powerful year. The tool was used by more platforms and identified more suspected CSAM than ever before, representing greater opportunities to remove abuse material and stop the cycle of harm for child victims.

5,303,413

Total suspected CSAM files found

3,844,868
Files classified as suspected CSAM

1,458,545
CSAM files matched

318,593,302

Total lines of text processed

1,369,024
Instances of suspected child exploitation found

 

86

Number of companies using Thorn’s detection products

415,375,895,003

Files processed by our customers

Human in the Loop

A webinar on the future of AI moderation

As part of Thorn’s expertise in platform safety, we wanted to continue a conversation that started at the 2025 TrustCon, an annual conference for trust and safety professionals. The advancement of AI and large language model capabilities has directly impacted platform moderation, with automation becoming an increasingly important and useful tool for trust and safety teams. We believe technology can be a force for good, but the discussion about how humans are involved when AI technology is employed in content moderation deserved a deeper dive.

We were happy to bring together four experts in the trust and safety space to explore how to build ethical AI for content moderation. 

Thorn in Action

Every year, we interact with the child safety ecosystem to educate, learn from, and partner with leaders in the space. These engagements give us the opportunity to advance important conversations, raise awareness, and share insights that make a difference in the lives of young people everywhere. By giving our subject matter experts visibility, we elevate the discussion around the threats children face and the global epidemic of child sexual abuse and exploitation.

TrustCon

In 2025, Thorn participated in four panels and had 14 employees attend the premier conference for people responsible for protecting users from harmful content. Our hosted panels surfaced important topics in the space:

  • Technology Designed With Youth in Mind
  • Ethical Approaches to Using AI in Content Moderation
  • Emerging Trends in Commercial Exploitation of Children

At one of the industry’s most visible events, Thorn was a major contributor of valuable, data-driven insight. 

The AI Conference

Dr. Rebecca Portnoff, Thorn’s VP of Data Science and AI, spoke at The AI Conference in 2025. This emerging event gave her a chance to present Thorn’s Safety by Design for Generative AI principles to an engaged and influential audience. Keynote opportunities like these are key to educating technology companies on best practices for building child safety into their platforms. 

Raising awareness through media engagements

Connecting the general public to the realities of child harm requires collaborative partnerships with respected media outlets. Thorn generates data, insights, stories, and resource guides relevant to caregivers across the nation, helping them better protect the young people in their lives.

We couldn’t do this work without you

Every insight, data point, and technological advancement is made possible by the support of the Thorn Cause Community. From our Builders who give month after month, to our donors who give annually, every gift helps move us forward as we transform how children are protected from abuse and exploitation in the digital age.

“As a father, it breaks my heart that there are children in harm’s way. I’m a Builder because every kid deserves to grow up in safety.”

Michael, Nebraska

“Who doesn’t want to protect children from online sexual abuse?  Who doesn’t want to end the exploitation of kids?”

Cameron, Italy

“I’m a survivor, and I will do everything I can to make sure no child suffers this kind of abuse. Together as Thorn Builders, we are making a difference.”

Lisa, North Carolina

Donor Spotlight: Jen Kao and Adam Leibsohn

The Thorn Cause Community is made up of supporters from all walks of life. Jen Kao ran a women’s wear label for over a decade. Adam Leibsohn, co-founder of GIPHY, a search engine for animated GIFs, has an extensive technology background. Their shared passion for Thorn’s mission is a great example of how people from all backgrounds and lived experiences can unite for child safety. Learn more about what brought Jen and Adam into the Thorn Cause Community, and why they support the important work Thorn does. 

When you donate to Thorn, you’re helping to fund a better tomorrow for all kids. Every dollar helps us conduct original research, build cutting-edge technology, and equip investigators to protect more children from sexual abuse.

Donate now

2025 FINANCIALS

Financial transparency

Your support enables us to make a difference for children every day.

Our unique dual-funding model provides a combination of philanthropic and earned revenue to support our child safety pillars:

  • Research & Insights: 100% philanthropically funded
  • Technical Innovation: 100% philanthropically funded
  • Child Victim Identification: 34% philanthropically funded; 66% earned revenue funded
  • Platform Safety: 100% earned revenue funded

Financial data is unaudited.