How do we know what we see online is real when AI-generated images, audio and video are becoming increasingly convincing?
There was a time when verifying a photograph or video meant getting hands-on with the evidence: checking metadata, inspecting images and cross-referencing details against satellite imagery or other sources. But as social platforms strip away useful information and AI-generated media becomes harder to distinguish from the real thing, the job of verifying content is changing fast.
In Episode 24 of The Signal, Tom Trewinnard and Fergus Bell explore whether new technologies can help us keep up. They look at Google’s SynthID, which embeds digital watermarks into AI-generated content designed to remain detectable even after cropping, screenshots and sharing, as well as other approaches to content provenance and verification. But what does it really mean for something to be labelled synthetic, and is watermarking enough to help us navigate an internet where seeing is no longer believing?
The conversation also turns to the wider media industry, where established organisations are under pressure to change how they work. At the BBC, there’s a call for a little less conversation and a little more action, while prominent journalists are exploring opportunities beyond traditional newsrooms and building new relationships with audiences.
And, as if that weren’t enough to keep up with, Tom and Fergus catch up on the latest developments in the AI model raceand what they could mean for the media.
From proving what’s real online to figuring out what journalism looks like next, this episode explores the technology and industry shifts reshaping the way we make, share and trust information.
Subscribe to The Signal for more conversations about AI, journalism and the future of media.
","author_name":"Fergus Bell, Tom Trewinnard"}