Meta’s AI Sends “Junk” Tips to DOJ, Prompting US Child Abuse Investigators’ Alarm

Meta’s AI‑driven moderation tools are now generating “thousands of tips each month” that law‑enforcement officials say lack the actionable detail needed to pursue child‑exploitation cases, according to testimony from IC ICAC special agent Benjamin Zwiebel before a New Mexico state court. Zwiebel described the influx as “junk,” noting that the volume of reports from Instagram, Facebook and WhatsApp has doubled between 2024 and 2025, yet many submissions omit critical evidence such as images, videos or full text excerpts. An anonymous IC ICAC officer added that the reports often contain only a vague indication that a crime may have occurred, leaving investigators without the identifiers required to locate perpetrators.

The IC ICAC task force, a nationwide network coordinated with the U.S. Department of Justice, relies on “cybertips” to prioritize investigations. However, the surge in low‑quality tips is straining resources, the officers warned. “We get a lot of tips from Meta that are just kind of junk,” Zwiebel said, emphasizing that the sheer number of submissions is “overwhelming” and that the lack of usable data “weighs on you to know that this crime occurred, but we can’t identify the perpetrator.” The problem appears most acute on Instagram, where the anonymous officer said the “information not being provided” has “skyrocketed” in recent months.

Meta pushed back, citing its long‑standing cooperation with law‑enforcement agencies. In a statement, the company highlighted that it handled over 9,000 emergency requests from U.S. authorities in 2024, resolving them on average within 67 minutes, and that it routinely reports child‑sexual‑exploitation imagery to the National Center for Missing & Exploited Children (NCMEC). Meta also pointed to its “improved tip reporting process,” which it claims helps NCMEC prioritize urgent cases and has been praised by both the DOJ and NCMEC for speed and accuracy. The spokesperson noted that Zwiebel himself recommended Meta’s teen‑account feature during his testimony, suggesting that the tool is currently the only viable safeguard for minors on the platform.

New Mexico Attorney General Raúl Torrez, who is leading the state’s lawsuit against Meta, acknowledged the company’s contributions while maintaining that the platforms still prioritize profit over child safety. Torrez said, “I do want to credit some of the social media applications and platforms, including Meta… they do report images to the National Center for Missing & Exploited Children.” Nonetheless, filings released in the case reveal internal alarms among Meta executives about the firm’s capacity to police child sexual abuse, indicating that the “junk” tip issue is not merely a procedural hiccup but a systemic challenge that could affect the broader U.S. enforcement ecosystem.

Industry observers note that the dilemma underscores a tension between automated moderation at scale and the need for high‑quality intelligence in criminal investigations. While Meta’s AI can flag vast amounts of potentially illicit content, the current workflow appears to generate a high false‑positive rate, forcing investigators to sift through a deluge of incomplete reports. If the trend continues, the IC ICAC task force may need to allocate additional resources to filter and verify tips, potentially diverting attention from other pressing cyber‑crimes. The situation also raises questions about how tech firms can balance rapid, AI‑driven detection with the precision required by law‑enforcement partners, a balance that remains unresolved as the litigation proceeds.