Bluesky users are in revolt after co-founder Jay Graber announced Attie, an AI-powered app that lets users create custom feeds through natural language prompts. The backlash is particularly fierce because Bluesky explicitly promised in November 2024 that it had "no intention" of scraping user posts for AI training — a clear dig at X's recent policy changes. Interim CEO Toni Schneider insists Attie is a "separate app" built on Bluesky's Atmosphere protocol, not integrated into the main platform.

This isn't just another AI product launch gone wrong — it's a fundamental breach of trust with a user base that explicitly migrated to Bluesky to escape AI integration. The platform positioned itself as an anti-AI sanctuary when AI backlash was hitting fever pitch across social media. Now they're asking the same users who fled X over AI policies to embrace an AI tool for customizing their feeds. Graber's defensive responses and resharing of posts calling AI critics "shortsighted" only amplified the anger.

The user responses tell the story: "Cool! How do we block it?" and "Thanks, we're good, no need to explain it further" dominated replies to Graber's announcement. The company seems genuinely surprised by the hostility, with Schneider emphasizing this is "people-focused" AI that serves users, not platforms. But that messaging rings hollow to users who chose Bluesky specifically to avoid AI entirely.

For developers building on social protocols, this is a masterclass in reading your room wrong. Bluesky's technical approach — keeping the AI separate from the core app — is actually smart. But launching any AI product to an explicitly anti-AI audience without extensive community consultation was always going to backfire. The lesson: your technical architecture doesn't matter if you've fundamentally misunderstood your users' values.