Every Child Safety Report Is a Handoff

This August, my colleague Sam Freeman and I attended the 38th annual Crimes Against Children Conference in Dallas, Texas.

At Cinder, I usually see child safety reporting from the platform side. We build the workflows teams use to review child safety cases and submit CyberTipline reports to the National Center for Missing & Exploited Children (NCMEC). What I do not usually see is what happens after a report leaves a platform.

CACC gave us a clearer view of the people on the other end: more than 4,500 detectives, prosecutors, forensic interviewers, pediatricians, victim advocates, analysts, and others working the same problem from different angles.

That mix matters. A case that starts as a flag may pass through a reviewer, a report, an NCMEC analyst, a detective, and eventually a prosecutor. Most of us only see one part of that path. The conference was a reminder that every handoff changes what is possible for the next person.

The report is not the finish line

Two years ago, we wrote about building Cinder's NCMEC integration. At the time, a lot of the work was mechanical: translating platform data into NCMEC's XML schema, pre-filling company details that would otherwise be typed into every report, and making sure media could reach NCMEC without Cinder ever storing a copy of it.

That work still matters. Platforms using Cinder submit through NCMEC's API instead of the web form. Company metadata is filled automatically. Media streams directly from the platform's own CDN to NCMEC, is hash-verified in transit, and is never written to our systems.

But CACC made something obvious that is easy to miss from the platform side: a report can be fast, complete, and well-formed, and still fail to tell an investigator what they need to know.

What makes a report usable

Context changes the shape of a case. A single flagged image can tell an investigator very little. That same image, submitted with the account behind it, the surrounding conversation, and related content, starts to show a pattern. Splitting one incident across many separate reports can hide the very connections that make a case understandable or prosecutable.

Location data can determine whether anything happens next. IP and geolocation often decide whether a report can be routed to the right jurisdiction at all.

Some fields carry legal weight. Details about whether and how a platform viewed content affect what investigators are permitted to do with it. A box left unchecked can stop someone from opening a file.

And classification depends on context. Material that looks benign on its own can mean something very different in the context of grooming or a known series. This is why human reviewers, and the context they add at the moment of decision, remain essential.

What we are building toward

Cinder now sends more than 10,000 reports every week on behalf of our customers. This year, we rebuilt NCMEC reporting on top of Cinder's workflow engine, so reporting configuration lives in reusable templates instead of being duplicated inside every workflow.

That change matters because child safety reporting is not one-size-fits-all. A platform can keep separate templates for different incident types, map its own data model onto CyberTipline fields once, and carry the right industry classifications, file annotations, incident details, reported-person information, and media metadata into every relevant report.

Reviewers can also add context at the moment they make a decision, at the report, media, and user level. Related media tied to the same account can travel in one report when a reviewer marks it as violative. One workflow decision can create the report, quarantine the content, and enforce on the account, instead of asking a team to coordinate those steps manually.

We have also kept access narrow by design. Viewing reported material in Cinder requires an explicit NCMEC admin role. General admin access is not enough.

What I took back

I left CACC with a longer list of ways our reports can become more useful to the people who receive them.

Protecting children online is an uphill fight. Offenders are persistent, and every new platform gives them another way to reach children. But the most hopeful part of the week was seeing how many people are working the problem from every angle: detectives, prosecutors, physicians, advocates, analysts, engineers, and platform teams.

This work is difficult. It is also not being done alone.