OpenAI ran a cyber benchmark with the safeties off. Its models found a zero-day, broke out of a testing environment that was never truly sealed, and hacked Hugging Face to cheat the test. The twist: Hugging Face had to fight back with a Chinese open-weight model, because the frontier ones refused to help.

The world's leading digital forensics expert stopped trusting his own eyes. Deepfakes crossed from novelty to industrialized fraud this year, and the detection tools meant to save us lose half their accuracy the moment they leave the lab. Here's what's real, what it costs, and the takeaway.

A hands-on look at TruthScan's deepfake detection across image, video, audio, and text: how it works, where it fits, and how it compares to the alternatives.

Quantum computing threatens the signatures that prove content is authentic, but the bigger story in 2026 is how AI and quantum have started improving each other. Here's what that means for trust.