AI deepfakes are testing HR’s ability to spot fake evidence

Diamond has already encountered the problem firsthand. “I did have a situation with a workplace investigation not that long ago where there were falsified communications made,” she said. “I don’t know whether the person used AI to create that, but he’d used something.”

The detection challenge she describes is backed by recent research. When identity verification firm iProov tested 2,000 U.S. and U.K. consumers in 2025, only 0.1% correctly identified every real and fake image and video sample, even after being told to look for fakes. And the FBI’s Internet Crime Complaint Center logged approximately $893 million in AI-enabled fraud losses in 2025, the first year the agency tracked AI as its own category.

When the evidence itself is on trial

Deepfakes create a two-sided problem for workplace investigations. Fabricated evidence can be submitted to support a false complaint, while employees accused of genuine misconduct can claim authentic evidence was manipulated. Either way, Diamond said the answer lies in old-fashioned investigative rigor.

“They should follow proper investigatory techniques in terms of seeing what they could find to corroborate the story,” she said. “For example, if someone is saying that an employee sent them a harassing text message, checking the employee’s phone as well as the alleged harasser’s phone to see if the messages are on both phones, checking the deleted files, performing forensics. Do the best they can to corroborate the story.”

When the stakes are high, she recommends bringing in outside help.