AI email scam losses hit $224K per case in new FBI data study

Simmrin Law Group reviewed FBI 2025 Internet Crime Complaint Center data on AI-related offenses and determined that business email compromise produced average losses of $224,123 per complaint. This figure exceeded the $145,097 average for AI-related investment complaints by 54.5 percent. BEC accounted for only 135 of the 22,364 total AI-related complaints yet generated $30.26 million in reported losses.

The 2025 financial impact table for AI-related crime categories lists the following with all headers and units preserved: AI-Related Crime Category, Complaints, Reported Losses, Share of AI Losses, Losses per Complaint. Business Email Compromise recorded 135 complaints, $30,256,592 reported losses, 3.4 percent share, and $224,123 per complaint. Investment recorded 4,356 complaints, $632,041,188 reported losses, 70.7 percent share, and $145,097 per complaint. Tech/Customer Support recorded 574 complaints, $19,457,078 reported losses, 2.2 percent share, and $33,897 per complaint. Confidence/Romance recorded 626 complaints, $19,041,653 reported losses, 2.1 percent share, and $30,418 per complaint. Government Impersonation recorded 260 complaints, $7,061,628 reported losses, 0.8 percent share, and $27,160 per complaint. Real Estate recorded 115 complaints, $2,699,085 reported losses, 0.3 percent share, and $23,470 per complaint. Other recorded 504 complaints, $11,750,591 reported losses, 1.3 percent share, and $23,315 per complaint. Employment recorded 691 complaints, $12,550,185 reported losses, 1.4 percent share, and $18,162 per complaint. Personal Data Breach recorded 1,204 complaints, $18,767,964 reported losses, 2.1 percent share, and $15,588 per complaint. Phishing/Spoofing recorded 803 complaints, $10,283,732 reported losses, 1.2 percent share, and $12,807 per complaint. National AI-related total recorded 22,364 complaints, $893,346,472 reported losses, 100 percent share, and $39,946 per complaint.

Investment fraud accounted for 70.7 percent of all AI-related losses according to the Simmrin Law Group study. AI-related BEC produced the highest loss per complaint while representing only 3.4 percent of total AI-related losses.

The Simmrin Law Group analysis reported that BEC losses per complaint reached 17.5 times the level for phishing and spoofing and 6.6 times the level for tech support fraud.

Non-AI BEC complaints nationwide reached 24,768 in 2025 with $3.05 billion in losses. The 2024 totals stood at 21,442 complaints and $2.77 billion in losses, reflecting a 10 percent rise in reported losses and a 15.5 percent increase in complaint volume.

BEC losses across all methods rose from approximately $2.95 billion in 2023 with 21,489 complaints to $3.05 billion in 2025 with 24,768 complaints.

The FBI defines BEC as a scheme in which criminals manipulate legitimate business communications to obtain unauthorized payments, with requests potentially originating from compromised accounts, impersonated executives, altered vendor instructions, or lookalike email addresses. The Simmrin Law Group study observed that AI-generated language can render such messages more convincing, yet the 2025 complaint data does not identify specific techniques employed in individual cases.

The FBI advises businesses to verify any payment instruction changes through trusted contact information rather than relying solely on the requesting email. This step holds particular importance for companies that handle large transfers.

The Simmrin Law Group analysis of FBI state reports found Wisconsin recorded the highest AI-related crime report rate at 21.30 per 100,000 residents, followed by Florida at 9.30 per 100,000 and Nevada at 8.38 per 100,000. California ranked eighth at 7.08 reports per 100,000 residents.

Per-capita financial loss rates placed the District of Columbia first at $822,324 per 100,000 residents, Connecticut second at $437,801 per 100,000 residents, and California third at $403,985 per 100,000 residents. The national average stood at $232,818 per 100,000 residents.

California recorded the largest raw totals with 2,787 AI-related crime reports and $158,989,594 in reported losses. The state average reached $57,047 per report.

NCMEC received more than 400,000 reports involving harmful AI-related content in 2025, including over 182,000 reports concerning possession, generation, or attempted generation of AI-related child sexual abuse material. Since tracking began in 2023 the center has identified at least 275 victims and more than 158,000 images and videos.

James Strahler II received a 15-year sentence after pleading guilty to charges that included cyberstalking, obscene visual representations of child sexual abuse, and publication of digital forgeries affecting at least six adult female victims and involving AI-generated material with children. UNODC has noted AI-assisted recruitment through misleading employment offers, fake identities, translation services, automated communications, and remote operational support in trafficking cases.

Police confirmed that the perpetrator of the January 1, 2025 cybertruck explosion in Las Vegas used generative AI to prepare the attack, which caused minor injuries to seven people. A March 2026 Federal Judicial Center survey of 929 judges found that 15 respondents, or 1.6 percent, had encountered challenges involving allegations that audiovisual evidence constituted a deepfake.

A May 2026 federal evidence committee report concluded that an amendment to Rule 901 specifically addressing deepfakes was not warranted at that time. California Evidence Code §1401 requires authentication of material before it can be received in evidence.