Driving industry impact
With over a billion children using our members’ platforms, every safety measure matters. While prevention is difficult to quantify, we track the actions that reduce risk and strengthen protections.
Measuring progress
The Tech Coalition’s voluntary mileposts help measure, track & drive industry’s impact on child safety.
The most effective safeguards prevent abuse before it happens—making their impact inherently difficult to measure.
The Tech Coalition helps make progress visible by measuring the tangible steps members are taking to reduce risk, close safety gaps, and strengthen protections.
Our voluntary milepost framework allows us to assess and benchmark progress. Mileposts are categorized and presented across three areas: prevention, detection & response.
Child-safety mileposts progressed by our members 2024 – 2025
Prevention mileposts
Detection mileposts
Response mileposts
Milepost progress in 2025
In 2025, member companies advanced 317 mileposts—more than double 2024. This represents an average of 5+ improvements per company, also doubling the pace of progress.
Progress is broadly distributed across all three areas, reflecting continued strengthening of safeguards and earlier integration of safety into product design.
Metrics beyond mileposts
In addition to mileposts, members report a range of indicators through our annual survey. These provide further insight into how companies are strengthening their capabilities to prevent, detect, and respond to online child sexual abuse.
Explore industry action across key child safety indicators below, and find more detail in our 2025 transparency report.
Prevention metrics
Prevention practices are evolving as companies embed safety-by-design into product development, alongside policies and solutions to deter harmful behavior.
Safety-by-design
Members use of safety-by-design approaches
Members are increasingly embedding safety-by-design in product development to identify and mitigate risks earlier, before harm occurs.
In 2025, 49 of 57 members reported using formal safety-by-design reviews, integrating safety into product development.
Earlier risk identification is also strengthening, with 37 members using structured risk assessment frameworks and 33 documenting OCSEA-related risks.
More advanced techniques are emerging, including red teaming (25 members), alongside safeguards such as age-based restrictions (31) and parental controls or safety tools (25).
Together, these indicators show that prevention is becoming more systematic, with companies identifying and addressing risks earlier in the product lifecycle.
Deterrence measures
Members use of deterrence measures
Platforms are strengthening deterrence through clearer policies, targeted in-product warnings, and expanded prevention messaging to discourage harmful behavior.
In 2025, 50 members reported publicly outlining policies and enforcement actions, setting clear expectations for users about unacceptable behavior and consequences.
Companies are also implementing in-product deterrence, including warnings for users at risk of harmful behavior (17 members) and messages alerting potential victims to risky interactions (10 members).
These approaches reflect growing recognition that prevention requires both discouraging harmful conduct and helping users recognize risk.
Age assurance
Members use of age assurance
Age assurance is being adopted by more companies as part of efforts to support age-appropriate experiences and strengthen child safety online.
Self-declaration remains most common (42 members), while higher-confidence methods are expanding, including document verification (26) and facial analysis (20).
Many companies now combine multiple signals, adapting approaches based on platform design and risk.
Detection metrics
Hashing remains a core input within a broader detection system, alongside layered signals and AI-supported tools that help platforms identify and triage emerging abuse.
Image hashing adoption
Members use of image hashing technologies
Industry adoption of image hashing technologies continues to expand, strengthening detection and removal of known CSAM.
In 2025, 49 of 57 members reported using formal safety-by-design reviews, integrating safety into product development.
Platforms are increasingly combining multiple tools to detect known CSAM at scale, using hashing technologies that match digital fingerprints of previously identified content.
In 2025, 37 members reported using PhotoDNA, making it the most widely adopted solution. Through the Tech Coalition’s PhotoDNA sublicensing program, companies can access and implement this technology more quickly.
Adoption of additional tools is also expanding, with 21 members using Thorn’s Safer Hash and 13 using PDQ. Many companies deploy multiple tools, with 17 members reporting internal hashing systems to strengthen detection across services.
Video hashing adoption
Members use of video hashing technologies
Video hashing adoption is accelerating as platforms strengthen detection of known CSAM in video content.
In 2025, members reported deploying video hashing technologies, including Thorn’s Safer Hash (15 members), Google’s CSAI Match (12), and PhotoDNA for Video (10).
Many companies combine multiple approaches, with some using shared tools alongside internal systems (9 members) to improve detection across formats.
Adoption has accelerated, with companies reporting no video hashing capabilities falling from 17 in 2024 to 13 in 2025.
At the same time, limited uptake of TMK+PDQF reflects industry learning.
As adoption grows, video hashing is becoming an integral safeguard, complementing established image-based detection.
Detecting OCSEA with classifiers
Members use of classifiers
Classifier-based detection is expanding rapidly, helping platforms identify previously unseen abuse across images, video, and text.
Platforms are expanding the use of machine learning and AI models to identify abusive material, complementing existing tools such as hashing and user reporting.
These systems analyze patterns in images, video, and text to detect potential abuse signals.
Over the past three years, adoption has grown significantly, with image classifiers rising from 13 members in 2022 to 41 in 2025, and video classifiers from 4 to 32.
Members are also applying classifiers to detect harmful behaviors, including CSA text (18 members), grooming (12), and sextortion (8), reflecting efforts to identify more abuse.
Response metrics
Companies are strengthening how they act on detected abuse, improving investigations, intelligence sharing, and transparency across the ecosystem.
Post-investigation enforcement tools
Member use of enforcement tools
Investigative responses are becoming more structured and consistent across the industry.
Account suspensions and content removal have become standard enforcement tools following OCSEA investigations. Nearly all reporting members use account suspensions or blocks, with content removal or visibility restrictions close behind, indicating broad alignment on core response measures.
Beyond these baseline actions, many companies are deploying feature restrictions, network or device limitations, and account re-verification. See the graph on member use of enforcement tools.
These measures allow platforms to intervene more precisely, limiting an offender’s ability to re-engage or escalate harmful behavior.
Across the industry, investigative responses reflect a maturing approach to enforcement, with companies applying tiered responses that combine multiple tools depending on the severity and context of abuse.
Together, these developments show that investigative responses are becoming more standardized, operationally structured, and better equipped to disrupt abusive activity.
Reporting to authorities
Types of reporting used by members
Members are modernizing reporting to authorities with faster, more standardized processes.
In 2025, 38 members reported OCSEA cases to authorities using both API and manual reporting, while 9 reported exclusively via APIs, reflecting growing adoption of more efficient mechanisms.
This shift indicates increased investment in systems that support faster, more consistent, and scalable reporting. Many members are also overhauling internal processes to improve how reports are generated, validated, and shared with authorities such as NCMEC.
While manual reporting remains in use, several companies are phasing it out as they transition to automated systems. Together, these developments reflect broader efforts to strengthen operational response and ensure actionable information reaches authorities quickly and consistently.
Members are also expanding supplemental reporting to provide authorities with richer, more actionable information for investigations.
Members providing supplemental reports
Transparency & accountability
Members aligned to Trust Framework
More members are aligning reporting with the Trust framework, strengthening transparency across the tech sector.
In 2025, 25 members were fully aligned with the Trust Framework and 17 partially aligned, reflecting growth in transparency reporting.
Companies also strengthened reporting in response to regulations such as the EU Digital Services Act, while maintaining alignment with Tech Coalition standards.
This expansion reflects broader efforts to provide clearer insight into how platforms address abuse.
To support this, the Tech Coalition committed to develop draft updates and guidance on prevalence measurement and generative AI risks, helping improve consistency in reporting, with finalization planned for 2026.