Deezer reports 44% of daily uploads are AI-generated music

Deezer's latest data shows AI-generated music has become a significant portion of content uploaded to streaming platforms. The company reported receiving almost 75,000 AI-generated tracks per day and more than two million per month, representing 44% of all new music uploaded daily.
Key details from Deezer's announcement
The consumption of AI-generated music on Deezer remains low at 1-3% of total streams. However, 85% of these streams are detected as fraudulent and demonetized by the company.
Deezer's AI-music detection tool, first launched in January 2025, has tracked significant growth:
- January 2025: 10,000 AI tracks per day
- September 2025: 30,000 AI tracks per day
- November 2025: 50,000 AI tracks per day
- January 2026: 60,000 AI tracks per day
- April 2026: 75,000 AI tracks per day
Platform measures against AI-generated content
Deezer implemented several technical measures to handle AI-generated music:
- Songs tagged as AI-generated are automatically removed from algorithmic recommendations
- AI tracks are not included in editorial playlists
- The company will no longer store hi-res versions of AI tracks
- Over 13.4 million AI tracks were tagged on the platform throughout 2025
Deezer started tagging AI tracks at the platform level in June 2025, becoming the first streaming platform to implement this system. French streaming service Qobuz announced similar plans in February 2026.
User perception and industry context
A November 2025 survey conducted by Deezer found that 97% of participants couldn't tell the difference between fully AI-generated music and human-made music. The same survey revealed:
- 52% of respondents said 100% AI-generated songs shouldn't be included in charts alongside human-made songs
- 80% said 100% AI-generated music should be clearly labeled for listeners
The announcement comes as an AI-generated track recently topped iTunes charts in the United States, United Kingdom, France, Canada, and New Zealand.
Other major streaming services like Spotify and Apple Music take different approaches, often combining filters to identify low-quality AI music with transparency efforts left to distributors.
📖 Read the full source: HN AI Agents
👀 See Also

Claude Code Telegram Plugin Bug: MCP Notifications Silently Dropped — Workaround via File Polling and tmux Injection
A Telegram plugin for Claude Code works correctly but inbound messages are silently dropped because Claude Code discards MCP notifications on stdio transport. A workaround uses file polling and tmux send-keys with ~5-9s latency.

NVIDIA Vera CPU Launched for Agentic AI Workloads
NVIDIA has launched the Vera CPU, a processor designed specifically for agentic AI and reinforcement learning workloads, claiming 50% faster performance and twice the efficiency compared to traditional rack-scale CPUs.

AI Agent Behavior Governance Gap Exposed by Summer Yue Email Incident
Meta's AI alignment director Summer Yue connected OpenClaw to her work inbox, and the agent deleted over 200 emails due to context compression mid-task, forgetting safety instructions. Current solutions focus on capability restrictions rather than real-time behavior evaluation.

OpenClaw Client Adds Cost Tracking and Per-Agent Spending Limits
New release adds spending caps per agent, live usage UI with circular progress bar, sub-agent management, skill toggling, and per-agent model selection.