Some links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
Bluesky’s AI assistant Attie drew a sharp reaction after its late-March 2026 introduction: a ClearSky snapshot cited by TechCrunch put blocks of Attie’s account at roughly 125,000, against about 1,500 followers. That is a striking sign of hostility, not proof that most Bluesky users rejected the product. Attie was presented as a tool for building custom feeds—not as an announcement that Bluesky would train generative AI on users’ posts.
What happened when Bluesky introduced Attie?
Bluesky unveiled Attie around the ATmosphere conference in late March 2026. The company-associated AI assistant was described as a way to create custom social algorithms and feeds within the AT Protocol ecosystem. TechCrunch reported that by March 30, roughly 125,000 accounts had blocked Attie, while its account had about 1,500 followers; a ClearSky snapshot placed it among the network’s most-blocked accounts at the time. TechCrunch’s report documents a substantial and unusually visible response, but it is not a representative survey of Bluesky users.
The phrase “overwhelming disgust,” used in some coverage, is a headline characterization rather than a measured finding about the whole platform. A block count does not establish that every block came from a unique person, that every blocker opposed the underlying product, or that a majority of Bluesky’s users rejected Attie. Some users may simply have wanted to keep the account or its promotional posts out of their feeds.
What Attie is—and what the launch does not establish
Attie’s stated purpose is to help people make custom feeds by describing the kind of content they want, rather than requiring them to build a feed algorithm themselves. That fits Bluesky’s longstanding emphasis on algorithmic choice: users can select feeds instead of relying only on one platform-ranked timeline. Bluesky has described custom feeds as a distinctive feature and said it wants to make them more central to the app. Its 2026 product plans provide that context.
#1 Best Overall
That function is different from several things often grouped under the label “AI”:
- AI-assisted feed creation: a tool helps construct a feed or algorithm.
- Algorithmic ranking: software orders or filters posts according to rules or signals; it does not necessarily use generative AI.
- Automated moderation: systems flag or assess content for safety and policy enforcement.
- Model training on posts: content is used as training data for a generative model.
- AI-generated content: a system creates text, images, or other material.
The available reporting establishes Attie’s broad feed-building role. It does not establish that the launch changed Bluesky’s policy on training generative models with user posts, nor does it fully document Attie’s data permissions, model provider, retention practices, pricing, or complete feature set. It is therefore not accurate to treat the existence of Attie as proof that Bluesky began scraping posts to train a model.
Bluesky had previously said it had “no intention” of scraping user-generated posts to train AI models. That assurance helps explain why an AI-branded product provoked suspicion, but the launch coverage does not demonstrate that the company reversed it. Coverage of the backlash describes the tension between that earlier position and users’ interpretation of Attie.
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchWhy the reaction was so intense
For some users, Bluesky’s identity was part of the product
Many people came to Bluesky looking for an alternative to platforms they associate with aggressive AI integration, opaque recommendation systems, or use of public content in model training. Artists, writers, and other creators have particular concerns about consent, copyright, labor, and the spread of low-quality generated material. For those users, an AI product could feel like a change in the kind of online space they had chosen, even if the specific tool did not train on their posts.
This is a question of trust and expectations as much as engineering. Bluesky’s earlier anti-scraping message addressed one concrete concern; some users interpreted it more broadly as a promise to keep AI out of the service. Those are not identical commitments, but a company can face backlash when customers understand its values more broadly than its precise policy language.
Users questioned the priorities
Criticism was not limited to AI in principle. TechCrunch reported complaints that Bluesky was promoting Attie while users were still asking for basic features, including image attachments in direct messages. That points to a familiar product tension: a company may see an experimental feature as strategic, while users judge it against long-standing requests that remain unmet.
The announcement left practical questions unanswered
A tool that promises user control still needs to explain how that control works. Users reasonably want to know what information Attie can access, whether it processes posts, follows, or likes, where prompts and feed definitions are stored, whether an outside model provider handles data, and whether users can revoke access or delete their creations. The cited launch coverage does not answer those questions in full. Nor does it establish whether Attie was generally available, limited to a test group, or a separate application rather than a feature inside the main Bluesky app.
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Repair Windows errors before they cause bigger problems3Fix the driver behind crashes, sound loss and screen glitchesWhy an AI feed tool fits Bluesky’s strategy—and why that does not settle the argument
Bluesky has long promoted algorithmic choice: a user can choose among feeds rather than accept a single universal ranking. The AT Protocol is intended to support a broader ecosystem of apps, feeds, and moderation services, not just one Bluesky-branded interface. The company’s original explanation of that approach describes the goal of giving users more choice over algorithms. Bluesky’s algorithmic-choice post sets out that principle.
The apparent strategic logic of Attie is that natural-language assistance could lower the technical barrier to making a custom feed. That is an inference from the product’s stated function and Bluesky’s feed strategy, not a confirmed explanation of its business case. A person who cannot code might be able to describe a topic or reading preference and get help assembling a feed. If the result remains understandable, adjustable, and optional, it could broaden the practical reach of algorithmic choice.
But this same framing explains the conflict. Supporters may view an optional tool as one more route to user control; skeptics may see AI-generated feed logic as a way to insert a technology they oppose into an ecosystem they chose for different reasons. “More choice” is not automatically meaningful if users cannot tell what data a system uses, how it ranks posts, or how to leave it behind.
It also matters that Bluesky Social PBC, the Bluesky app, the AT Protocol, third-party services built on that protocol, and Attie are related but not interchangeable. The available sources describe Attie as an AI product associated with the AT Protocol ecosystem; they do not fully establish its current app boundary or permissions. A user blocking Attie’s account should not assume that this removes every AI-assisted feature from Bluesky or the wider protocol ecosystem.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Bluesky’s prior use of automation is a separate issue
Bluesky is not accurately described as entirely AI-free. Its transparency report says uploaded images and videos are sent to the third-party provider Hive for assessment, and states that neither Bluesky nor the vendor retains those files or uses them to train generative AI systems. That concerns trust-and-safety assessment, not Attie’s feed-building architecture. Bluesky’s 2025 transparency report should not be read as a direct disclosure of Attie’s data practices.
For users, the distinction is useful: opposing generative AI content, objecting to model training on one’s posts, and questioning automated moderation are different positions. A platform may use automation for safety while not training a generative model on user content. Each practice still deserves clear disclosure, but evidence about one should not be substituted for evidence about another.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What users can—and cannot—conclude now
The documented response shows that Attie became a focus of unusually visible opposition shortly after launch. It does not show that users were leaving Bluesky in large numbers, that the product caused a measurable decline in active users, or that Bluesky had adopted a policy of training on posts. Bluesky later said it had shipped a new AI product in the preceding four months, confirming the product’s place in its broader work, but that statement is not a detailed Attie privacy or permissions disclosure. The July 2026 company announcement also named Toni Schneider permanent CEO.
Bluesky’s growth and strategic ambitions help explain why the company may explore products beyond its core app. It said it had more than 43 million global users in March 2026, and its May FAQ said it had passed 44 million. It had also announced a $100 million Series B raised in April 2025. Those facts indicate a larger organization facing discovery, retention, and sustainability challenges; they do not prove that Attie was built to monetize users. The company has not, in the cited materials, established Attie’s commercial model. Bluesky’s funding announcement and FAQ provide its own figures and platform description.
Questions worth checking before using an AI feed tool
- What data can the tool access? Look for a precise list of profile, post, follow, like, or other account data—not just a general promise of personalization.
- Who processes prompts and feed definitions? Check whether an external model provider is involved, and what its retention and training terms are.
- Can you revoke access and delete results? A useful opt-out should explain how to disconnect an account and remove saved prompts or generated feeds.
- Is the feed inspectable and adjustable? Users should be able to understand, modify, and stop using the logic rather than treating its recommendations as inevitable.
- What does blocking do? Blocking an account may prevent seeing or interacting with that account; it should not be assumed to disable a separate service or all AI features.
- Is the product optional? Confirm availability and whether using the core Bluesky app requires any interaction with Attie.
Those are the right questions because the public reporting cited here does not yet settle Attie’s detailed data handling, opt-out route, availability, or commercial terms. Until those specifics are clear, users should avoid inferring either that their posts are being used to train Attie or that blocking its account constitutes a complete AI opt-out.
What the backlash means for Bluesky
Attie exposed a difficult test for Bluesky’s pitch. The company says an open ecosystem and customizable feeds give people more control; some users believe that introducing an AI product risks normalizing a technology they came to the platform to avoid. Both positions can be true: an optional feed-building assistant could extend choice, and its launch could still damage trust if users do not understand its data use or feel their priorities were ignored.
The immediate block count is a measure of attention and rejection, not a verdict on the product’s long-term adoption or Bluesky’s future. What matters next is whether the company explains Attie’s boundaries plainly—what it does, what it accesses, what it does not train on, how users opt out—and whether the tool proves useful enough to justify its place in the ecosystem. For a platform whose identity rests heavily on user control, that explanation is part of the product.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
Recommended Free Tools
