What type of crime is rising fastest and why?

Mainstream coverage this cycle points to identity theft/fraud as "America's fastest-growing crime," alongside felony assault climbing in NYC and a pandemic-era car-theft spike that's now reversing. The crowd's 30-day conversation doesn't dispute that framing — it overwhelmingly converges on it too — but adds a sharper, more specific mechanism the headlines gloss over: it's not "fraud" in general, it's AI-generated fraud, and specifically voice cloning, that people across platforms (mostly X, plus supporting web reports) treat as the actual driver of the surge.

What I learned: the crowd's core addition is that AI has collapsed the cost and skill needed to run identity/fraud scams at scale. Multiple posts cite the same anchor stats — the FBI logging AI-linked fraud losses around $893M in 2025 as its own tracked category for the first time[2][4], and a separate estimate putting global AI fraud losses at $442B[7] — with commenters framing 2025/2026 as the year AI fraud became a distinct, officially-recognized crime category rather than a subset of ordinary scams. IdentityIQ's report ties this to a 65% surge in fraudulent new-account activity, explicitly attributing it to AI making scams "appear more legitimate than ever"[1].

The specific mechanism the crowd keeps returning to, more than the top-line dollar figures, is voice cloning from minimal audio. Posts describe cloning a voice from "3 seconds of audio"[6] or "a few seconds" of social-media clips[2], used to power grandparent scams, vishing calls, and even fake job interviews[4]. One thread walks through how this defeats SIM-swap protections entirely: attackers no longer need to socially engineer a human at a telecom carrier — they just clone the victim's voice and "become" them to reset email, banking, and crypto access[6]. Security.org's grounding confirms this is technically feasible now — deepfake video from public photos, cloned voice from a 30-second sample, synthetic SSNs and fabricated credit histories combined into one fraud package[3] — which the crowd treats less as a hypothetical and more as the current baseline threat.

A secondary crowd thread applies this same lens outside the US: UK investment fraud reportedly jumped 40% in the past year, with investigators there also pointing to AI-generated websites and voice cloning as the scaling mechanism for organized fraud groups, not just lone scammers[8]. This global angle isn't really in the mainstream US coverage provided, which stays domestic (FTC identity theft stats, DC/NYC crime data, ITRC's "multi-layered" crimes report on minors and employment fraud).

Notably, the crowd evidence here is thin on other crime categories the mainstream baseline flags — there's no meaningful Reddit/TikTok/HN discussion in this dataset directly weighing in on felony assault trends, organized crime homicides, or the motor-vehicle-theft reversal; those remain effectively mainstream-only stories in this cycle. The crowd's attention is heavily concentrated on AI-enabled fraud/identity theft specifically, with X carrying the bulk of volume (55 items) and driving nearly all the distinct claims, while Reddit/TikTok/Instagram/HN activity in this sample didn't surface a competing or dissenting narrative — it's a case of the crowd narrowing and technically deepening one mainstream storyline (identity theft as fastest-growing crime) rather than challenging it or surfacing a different candidate crime entirely.