LANDR Says Spotify-Flagged Artificial Streams Fell 95% After Its Anti-Fraud Push

The fight against fake streaming is moving further upstream.

Instead of waiting for Spotify or another DSP to detect suspicious activity after it happens, distributors are increasingly trying to identify problems before music even reaches the platforms.

LANDR is one example.

The music creation and distribution company says artificial streams flagged by Spotify across music it distributes dropped by 95% between Q1 2025 and Q2 2026. LANDR announced the figure while expanding fraud-detection technology developed with Montreal-based AI research institute Mila.

There is an important distinction here: the figure comes from LANDR itself. It does not mean artificial streaming across Spotify as a whole has fallen by 95%.

LANDR also does not claim that one AI system caused the entire reduction. The company says several fraud-prevention measures were introduced over the period, with its Mila-developed machine-learning system forming one part of the wider strategy.

Distributors are checking more than audio files

Digital distribution used to feel relatively straightforward.

Upload a WAV, add the metadata, choose a release date and send the record to the stores.

That model is changing.

LANDR says its systems now look for signals including unusual release frequency, catalogues appearing very quickly, accounts sharing devices or payment information, and metadata that may be attempting to impersonate established artists.

One machine-learning model analyzes account and behavioural information, while another looks at listening activity for patterns that may not resemble a genuine audience.

Human review remains part of the process.

This is important because unusual activity does not automatically mean fraud. Sometimes a record simply takes off.

The bigger issue for independent artists

Most legitimate artists are not running bot networks themselves.

The risk often appears through marketing.

An artist pays a company for promotion. That company uses another provider. The release suddenly receives thousands of streams from playlists or accounts the artist has never seen before.

Then the distributor or DSP detects suspicious activity.

This is why phrases such as “guaranteed Spotify streams” should be treated very carefully.

A serious music marketer can guarantee a campaign, an advertising budget or a number of impressions.

They cannot guarantee that a specific number of real human beings will choose to stream a song.

Good numbers and useful numbers are not the same thing

Artificial streaming can make a dashboard look impressive while making the underlying artist data almost useless.

Where are the real listeners?

Which cities genuinely respond to the music?

Who came back for a second track?

Who saved the release?

Who followed the artist afterwards?

Those are the signals that can actually help a career.

For labels as well, cleaner data makes it easier to understand whether a campaign worked and where an artist is genuinely developing an audience.