Ed Sheeran can be heard singing a testimony of God’s redemption – in a false AI-generated song.
Almost indistinguishable from Sheeran’s ‘real’ voice, the fake sings:
“My feet were draggin' slow on this road, dark clouds above me heavy load // Caught in shadows cold and lone, then a whisper said you're not alone…
“Jesus lift my soul so high, break my chains let me fly //Through the storm through the fire seekin' love seekin' higher.”
The video features a series of videos stitched together, some real and some AI-generated.
Whilst the Sheeran video is seemingly light-hearted, the UK government is cracking down on deepfake AI.
Deepfakes are generated images, which use artificial intelligence to manipulate pictures of, or portray real people.
New offences will criminalise the creation and sharing of explicit deepfake images.
Justice Minister Alex Davies-Jones said perpetrators of the new offences would "face the full force of the law".
Where bending truth is concerned, the fake-Sheeran video seems to have YouTube users fooled. Comments include: “So good to see Ed Sheeran singing the name of the Lord of all Lords and King of the universe Jesus Christ our Saviour,” and, “Wow one from Justin B and now Ed Sheeran! God is making big moves!”
It also appears that many of the comments themselves are from artificial ‘bot’ accounts, which pose as real users. Hallmarks of bot accounts include long, numerical usernames and frequent poor grammar in comments.
James Poulter is a leading AI and innovation expert, and Christian. He told Premier Christian Radio: “There’s a real opportunity for the church to have a dialogue with the public around what it really means to be a human person, and what it means for AI to come alongside us and be a help and not a hindrance."
He told Premier that the issue does not stop with regulation being introduced – but hinges on who is being regulated. “We want this technology for the benefits, but we want to protect against the harms,” he said. “Where are we going to put that protection?
“The way things are created is obviously very hard to regulate, because many of these models are what's known as ‘open source’. That means that the code is out there for people to download it and run it on their own machines, and use it to develop products and projects out of.
“But we could regulate it on the distribution end, which is really the social networks the likes of YouTube and Meta's platforms, Tiktok and elsewhere.”