identifAI, a deepfake-detection firm established by Marco Ramilli and Marco Castaldo, has unveiled research into deepfake-driven fraud spanning from January 2020 through August 2026. The analysis reveals patterns in how synthetic media is weaponised across different platforms and geographies.

The fraud landscape according to identifAI

Video synthesis dominates the fraud incidents tracked by the company, representing 42.7% of cases, with voice cloning accounting for 21.9%. Combined, these two attack vectors constitute nearly two-thirds of all incidents at 64.6%.

  • The United States leads in incident frequency at 34.1%, followed by India at 7.7% and the United Kingdom at 6.6%
  • X emerges as the primary platform where such incidents surface, at 42.1% of cases, with LinkedIn trailing at 14.5%
  • Identity fraud represents 17.6% of deepfake incidents, though the company attributes its claim that synthetic media underlies roughly 1 in 100 failed identity checks to Identity Week rather than its own measurements

Understanding the methodology

The report's scope depends heavily on how identifAI defines what counts as an incident. The company classifies an incident as a deepfake deployed in real-world contexts that either receives coverage from English-language media outlets or appears in social media discussions. Classification relied on a large language model with subsequent human verification.

Out of 59,747 deepfake events observed, the fraud category encompasses 7,926 incidents. Notably, 55.8% of victims fall into an "other or unspecified" classification, reflecting gaps in available information.

The $40 billion figure featured in the press release's headline represents a projection of US fraud losses anticipated by 2027, derived from unnamed "financial think tanks" rather than direct measurement by identifAI.

The path forward

Ramilli contends that deepfake detection "can no longer exist as a post-incident forensic tool" and must instead be embedded into financial systems as an active security layer, suggesting a shift in how institutions should approach synthetic media threats.

Source: Startup Reporter