US and UK law enforcement are bracing for an explosion of AI-generated CSAM, and criticize Meta, the authorities' best partner for flagging CSAM, over E2EE
Eileen Sullivan / New York Times :
Context & Ripple Effects
Law-enforcement pressure on Meta's encrypted messaging strategy was already visible in a joint warning from international agencies, which argued that wider E2EE would reduce visibility into CSAM reporting. The present concern adds generative AI to that long-running detection-versus-privacy conflict.
Meta's role is unusually consequential because it supplied the overwhelming share of suspected-CSAM tips reported to NCMEC in 2022, even as prosecutors warned that high reporting volume can slow casework. AI-generated material raises both the expected volume and the difficulty of sorting actionable reports.
First-order effects
- US and UK agencies must prepare investigative and triage capacity for more suspected AI-generated CSAM, while continuing to depend heavily on Meta-originated reports.
- Meta faces renewed public and policy pressure over E2EE because encryption limits the platform's ability to identify and flag material in private messages.
Second-order effects
- The debate shifts from detection alone to operational quality: a larger flow of AI-related reports can force investigators and reporting platforms to prioritize signals that lead to actionable cases.
- Other messaging providers face stronger demands to explain how they will support child-safety enforcement without weakening private communications; prior EU scanning proposals show how quickly that pressure can become a regulatory question. EU plans for mandatory CSAM scanning made the same E2EE trade-off explicit.
Third-order effects
- If AI-generated abuse material expands as authorities expect, CSAM policy will increasingly test whether safety systems can scale alongside private, encrypted services rather than treating those objectives as separately solvable.
- The durable fault line is likely to be governance over detection, reporting, and access to private communications—not simply whether platforms deploy AI—though the balance will remain contested between child protection and privacy advocates.
The trend: Generative AI is intensifying the longstanding clash between platform privacy architecture and governments' demand for scalable online-safety enforcement.