aiqbench takeAI Safety

This research cluster shifts deepfake detection from adversarial pattern matching to verifying "physical anchors" inherent in real-world sensor data and environmental conditions. By focusing on manufacturing imperfections in camera sensors and electrical grid fluctuations, the approach aims to identify synthetic media that lacks these physical signatures. This strategy offers a robust alternative to traditional AI detectors that often struggle to keep pace with advancing generative models.

Source coveringGDELT

What this isA model's reading of the article and related coverage, written for aiqbench. Not the publisher's words.

Research Cluster Takes Real-World Approach to Detecting Deepfakes

GDELTOctober 1, 20265 min read

A new department-level research cluster at Binghamton University is betting that the answer to combating fake images and video generated by artificial intelligence is not better detectors, but real-world physics. On Thursday, the Department of Electrical and Computer Engineering at the Thomas J. Watson College of Engineering and Applied Science launched the Cyber-Physical AI Trust and Integrity (CAITI) research cluster. "Trust is becoming one of the defining challenges of the digital age, and the Department of Electrical and Computer Engineering at Binghamton University has been working on its physical foundations for decades, from sensor…

Excerpt shown under the publisher's RSS terms · 94 words · full text and figures are on the publisher's site

Read at GDELT ↗Ask about this article