Insights on verifying and detecting generative AI

A comprehensive report on the evolving landscape of synthetic media threats and enterprise security implications from Q2 2025.

When BTS Army heard the AI cover of “Romania,” many thought their idols had dropped a secret new track. The voices were so lifelike, the emotions so real, it felt like technology had found its own way to make music breathe. AI-generated singing with celebrity voices isn’t just im

Playing synthetic audio through a speaker and re-recording it, a "replay attack", can disguise deepfakes as genuine, raising a leading open-source detector's error rate from 4.7% to 18.2%. Research by Resemble AI's Nicolas Müller, Piotr Kawa, and Aditya Tirumala Bukkapatnam, presented at INTERSPEECH 2025.

When BTS Army heard the AI cover of “Romania,” many thought their idols had dropped a secret new track. The voices were so lifelike, the emotions so real, it felt like technology had found its own way to make music breathe. AI-generated singing with celebrity voices isn’t just im

In a striking example of the challenges posed by AI in the creative industry, two professional voice actors recently discovered unauthorized AI-generated replicas of their voices circulating online. Initially contracted to provide voice samples for what was described as internal

In California, an elderly man named Anthony received a chilling phone call. The voice on the line was his son’s, panicked, injured, and begging for help after a supposed car crash. It sounded real. It felt urgent. So Anthony wired $25,000 for bail money . Only later did he learn