Bluesky users pushed back hard this weekend against a new AI-powered feed tool, turning it into one of the most-blocked accounts on the fledgling social network — an early sign of anti-AI sentiment on a platform that markets itself as a decentralized alternative to Elon Musk’s X.
Attie, an experimental feed-building app launched publicly Saturday by The Atmosphere, was blocked about 125,000 times in the hours after its announcement, according to analytics site ClearSky. That places it second only to U.S. Vice President JD Vance — whose Bluesky account remains the platform’s most-blocked at 180,684 blocks — and ahead of high-profile government accounts like the White House and ICE, each blocked by more than 100,000 users.
Built on Bluesky’s AT Protocol, Attie uses AI to assemble personalized feeds: users type a short description of the posts or topics they want to see, and the tool searches across Bluesky to compile a custom stream. The Atmosphere team that created Attie is led by former Bluesky CEO Jay Graber, who now serves as Bluesky’s chief innovation officer.
The launch prompted swift backlash from a vocal segment of the community. Writers and creators posted snarky reactions — “It would be kinda neat if Attie became the most-blocked account,” author Dani Finn wrote — and others warned the move threatens what drew them to Bluesky in the first place. “Most of your user base came here because they wanted to get away from Twitter's AI right?” illustrator Marco Alfaro wrote, arguing the feature undermines one of Bluesky’s competitive advantages.
Some critics framed the rollout as a symptom of platform maturation. Tech YouTuber Sam Thibault suggested the shift reflects broader priorities that come with growth: building new market-facing features instead of focusing on persistent platform issues.
Bluesky’s open analytics helped make the surge visible — X does not publish similar block metrics publicly, so comparisons across platforms are opaque. Blocking and shared blocklists are already a core part of Bluesky’s culture and user-driven moderation, making blocks a frequent and highly visible form of pushback.
Graber acknowledged user concerns in comments to Decrypt: “We understand that some of our users have genuine concerns about how LLMs work and the impact they are having on our society. We take those concerns seriously.”
The episode underscores a broader tension for decentralized social apps: as they scale and integrate AI-driven features, they must balance experimentation and growth with a community that in many cases migrated specifically to avoid algorithmic personalization.
Editor’s note: this story was updated to include comment from Bluesky.
Read more AI-generated news on: undefined/news
