Deepfake fraud has produced an estimated $1.56 billion in cumulative losses since 2017, according to a Surfshark research chart built from the AI Incident Database and the Resemble.AI Deepfake Incident Database. The data runs through September 2025.

Losses have accelerated. Deepfake-related losses in 2025 alone exceeded $1 billion, up from roughly $400 million in 2024. The five years from 2019 through 2023 added up to about $130 million combined. Investment scams account for the largest share at $900 million, or 57 percent of the losses in the analyzed cases.

The cost of producing deepfakes has fallen as the losses grew. Video deepfakes that once required professional skills and cost between $300 and $20,000 per minute now cost a few dollars per minute. Synthetic voice generation costs between $0.01 and $0.20 per minute, and voice cloning needs only about three seconds of recorded audio.

The chart lists individual cases that show the range of harm. One voice scam victim lost $15,000. Lost-pet scams ask for payments that are sometimes as small as $50. In a hiring scheme, more than 100 companies unknowingly employed fraudulent remote workers.

Surfshark, a cybersecurity and privacy software company, compiled the figures from incident entries that the two databases logged between 2017 and September 2025, so later cases fall outside the total. The totals combine incidents from all countries and the chart gives no US-only breakdown.

Source: Surfshark - https://surfshark.com/research/chart/ai-deepfake-losses