Paul McCartney and Ringo Starr release Now and Then, the first “new” Beatles song since 1995, created using AI and an old lo-fi John Lennon recording
with help from AI James Farrell / SiliconANGLE : Beatles get back together for one last song with the help of artificial intelligence Maria Sherman / Associated Press : Listen to the last new Beatles' song with John, Paul, George, Ringo and AI tech Alexis Petridis / The Guardian : The Beatles: Now and Then review - ‘final’ song is a poignant act of closure William Gallagher / AppleInsider : Tim Cook highlights The Beatles' last song on Apple Music Mary Kate Carr / The A.V. Club : The Beatles' final song, “Now And Then” is here X: @thebeatles : #NowAndThen is now available to listen to worldwide! https://thebeatles.lnk.to/... [video] Paul McCartney / @paulmccartney : ‘Now And Then’ by @thebeatles is now available to listen to worldwide. https://thebeatles.lnk.to/... #NowAndThen [video] Tim Cook / @tim_cook : A song 40 years in the making — @thebeatles' ‘Now and Then’ is available on Apple Music today! https://music.apple.com/... Damon K / @dada_drummer : The power of AI seems to have delivered us a new Rutles track... Here's the original demo on youtube https://www.youtube.com/... @thebeatles : NOW LIVE! Watch the documentary and learn the full story behind ‘Now And Then’ - with exclusive footage and commentary from Paul, Ringo, George, Sean Ono Lennon and Peter Jackson. https://thebeatles.lnk.to/... #NowAndThen @johnlennon @PaulMcCartney @GeorgeHarrison @ringostarrmusic... [video] Tom Warren / @tomwarren : The Beatles' last song is now streaming thanks to AI. Machine learning helped Paul McCartney and Ringo Starr turn an old John Lennon demo into what's likely the band's final song https://www.theverge.com/... LinkedIn: Daniel Tunkelang : The track is listenable but strikes me as mediocre compared to the Beatles' work as a whole. It's what I'd expect from AI-generated content — (UPDATED) even though it's really humans to blame. Benjamin McCoy : In both lectures I gave this year, someone asked how AI would affect creatives. — AI is just tech. Tech will never replace art, but it can enhance it. … Shobha Lizaso : AI brings us a new Beatles song in 2023! It started out as muddled lo-fi recording of John and a piano from 1977. … Wendy O'Donnell / Wendy O'Donnell, MBA : One of the best things resulting from AI? Hearing a new Beatles collaboration in 2023✌️ Victor Prilutskiy : One of the greatest uses of tech progress. The Beatles' final song is now streaming thanks to AI. — https://lnkd.in/dkXQ5dsf … Sean Byrne : This morning, while I was on the subway to the office scrolling on my phone, I felt excited about The Beatles' latest song release. … Luciana Padua : When everyone is talking about the bad side of AI, it's really beautiful to hear this story. The interaction of technology and humans to make beautiful things. … Monserrat Soto Roig : 🎵 On this second attempt The Beatles finally did it, with a little help from friends and AI. … Robert Julian Smith : Machine learning helped Paul McCartney and Ringo Starr turn an old John Lennon demo into what's likely the band's last collaborative effort. ❤️ See also Mediagazer
Context & Ripple Effects
The release follows McCartney’s account that a sound editor had used AI to isolate Lennon’s voice from a demo, turning a technical restoration step into a finished Beatles recording.
It sits apart from systems such as OpenAI’s early raw-audio music generator: the reported use here was to recover a usable vocal from archival material, not to synthesize a new Lennon performance. Its later Grammy recognition reinforced that distinction.
First-order effects
- McCartney and Starr can complete and commercially release a recording built around previously unusable Lennon source material, extending the Beatles catalog with surviving members’ authorization.
- The project makes AI-assisted audio separation visible to audiences as a production tool for legacy recordings, while keeping the underlying creative inputs tied to identified artists and an existing demo.
Second-order effects
- Catalog owners and producers gain a prominent proof point for revisiting degraded or incomplete archives, increasing demand for restoration workflows that can separate vocals and instruments without replacing performers.
- The release sharpens the practical distinction between AI restoration and prompt-generated music—a distinction platforms and labels may need to communicate as generative-music services such as Suno’s text-to-song product emerge.
Third-order effects
- If more archival projects follow, AI may become a standard layer of catalog exploitation: less a substitute for artists than a way to unlock recordings previously constrained by source quality.
- The resulting market will put greater weight on provenance and disclosure, because audiences, rights holders, and distributors will need to distinguish recovered human performances from newly generated audio.
The trend: AI music is moving from experimental generation toward commercial workflows that restore, package, and monetize rights-cleared archival recordings.