Detecting AI deception in the real world: lessons learned from the experts

Descrizione

Synthetic media, AI and deepfakes continue to advance in their realism, ease of creation and accessibility. While AI was not the pivotal factor in global elections in 2024, significant audio, image and video deepfakes impacted major elections - while both the public and journalists wrestled with people in power using plausible claims that content was created with AI to dismiss genuine content.
Critical faultlines are emerging around how journalists, fact-checkers, civil society and election officials handle both actual deceptive AI and the use of claims of AI to obfuscate reality. And as multimodal generative AI is now the ‘worst it will ever be’ in terms of hyper-realism, and with the advent of increasingly realistic audio and video how can we best prepare journalists and the media globally?
This panel session will provide insights into how frontline journalists are grappling with detection of deepfakes and malicious AI, explore insights from the WITNESS Deepfakes Rapid Response Force and highlight how there are fundamental gaps in detection globally that need to be addressed to support journalists as well as discuss the state of the science with a leading technical researcher.