The Debrief
L7L14L30L90All
PaidSearchIndustryTechDataBrandConversion
Brand · 2 min read22 July 2026

Half of Deezer's Daily Uploads Are Now Made by a Machine. Trust Is the Only Moat Left.

Deezer says fully AI-generated tracks passed 50% of daily uploads at peak in June 2026, around 90,000 songs a day. It is a preview of every content category once production costs hit zero. When volume is infinite and free, the scarce thing becomes trust.

The cost of making content just fell to zero. The value of being trusted to recommend it just went up.

2 min read

For the first time, more than half the music uploaded to a major streaming service in a single day was made by a machine. Deezer says fully AI-generated tracks passed 50% of daily uploads at peak in June 2026, running at around 90,000 AI songs a day. In April the figure was 44%. The line is still climbing.

Deezer has been tagging AI music for listeners since June and running its own detection tool since January 2025. It now strips AI tracks from algorithmic and editorial recommendations, and it will remove AI songs used for streaming fraud or left unplayed for six months. The platform is drawing a line between content that exists and content that earns a place in front of a human.

This is not only a music story. It is a preview of what happens in every content category once the marginal cost of production hits nothing. The flood arrives, the platforms respond with detection and labelling, and the scarce thing becomes credibility, not volume.

Why it matters

For anyone producing content in Australia, the lesson is blunt. Volume stopped being a moat the day a machine could match it. If your content strategy is built on publishing more than the next brand, you are competing with an infinite supply that costs nothing to make. The brands that win from here are the ones a person or a platform actually trusts.

50%

The share of daily uploads on Deezer that were fully AI-generated at peak in June 2026

What to do about it

Stop measuring content by volume. Measure it by whether anyone trusts or acts on it.
Build signals of credibility that a machine cannot fake. Real expertise, real data, a real name behind it.
Expect labelling and detection to spread to your channels. Assume your content will be sorted into human or machine.
Put your effort into the pieces only you could make, not the ones anyone could generate.

The machines can make the content now. They still cannot make the trust. That is the whole business from here.

Share this brief
Send it to a colleague who'll find it useful.
Filip Ivanković
The Debrief / From Filip Ivanković
One every morning. Six months in, you'll see the patterns most don't.
Strategy, benchmarks, and what's actually moving in Australian marketing. Four-minute read. The reps compound.
Filip Ivanković·Founder, New RebellionAboutLinkedIn