Bluesky
PulseAugur coverage of Bluesky — every cluster mentioning Bluesky across labs, papers, and developer communities, ranked by signal.
- founded by Jack Dorsey 100%
- employed by Toni Schneider 95%
- partners with Jetstream 90%
- authored Emily M Bender 90%
- founded Bluesky Jetstream 90%
- authored Alex Hanna 90%
- uses AT Protocol 90%
- developed by Attie 90%
- authored by The AI Con 90%
- founded by Jay Graber 90%
- developed Attie 90%
- used by Jetstream 70%
- 2026-08-26 product_launch Bluesky enhanced its video upload capabilities by increasing the maximum video length and file size, and improving upload speeds. source
- 2026-08-11 product_launch Bluesky launched a feature allowing users to hide reposts from specific individuals. source
- 2026-08-06 partnership Bluesky's Jetstream is expanding its partnership with Hugging Face to accelerate open science. source
- 2026-07-11 hiring Toni Schneider has been officially named CEO of Bluesky, taking over from Jay Graber. source
- 2026-07-11 hiring Toni Schneider has been officially appointed as the CEO of Bluesky, transitioning from his interim role. source
- 2026-06-11 product_launch Bluesky announced it will launch a communities feature later this year, similar to Reddit's. source
- 2026-06-11 product_launch Bluesky is introducing a new 'communities' feature to create smaller, focused discussion spaces. source
22 day(s) with sentiment data
How is Bluesky enhancing its content moderation capabilities?
Bluesky is significantly advancing its content moderation through new AI benchmarks and detailed studies of its existing systems.
A new ModerationBench, built with 4,000 annotated Bluesky posts, shows foundation models nearly triple the F1 score of Bluesky's current system. Concurrently, an arXiv study of the Bluesky Moderation Service (BMS) reveals high precision but low recall in harm detection, highlighting areas for improvement in human-AI collaboration for complex issues like hostility towards protected groups.
What new AI tools are empowering Bluesky users and developers?
Bluesky is seeing a surge in AI-powered tools for user account management and streamlined software development.
The Attie AI assistant has launched on Bluesky, enabling personalized feed generation by integrating with models like Claude. Additionally, the open-source bluesky-mcp project allows AI assistants to manage Bluesky accounts, performing actions like posting and replying. For developers, SpecKit uses AI to generate software requirements before coding, as demonstrated on Bluesky.
How is Bluesky's Jetstream platform evolving for AI research?
Bluesky's Jetstream continues to be a dynamic environment for AI experimentation and open science, marked by key partnerships and new applications.
The expanded partnership with Hugging Face has tripled storage capacity and boosted download speeds for large datasets, supporting open science. Researchers like Ethan Mollick and Jürgen Schmidhuber continue to launch interactive AI web apps on Jetstream, showcasing its versatility for prompt testing and image generation, including a custom "Codex pet" created by Simon Willison.
What ethical and efficiency challenges are emerging for Bluesky's AI?
Bluesky is navigating critical discussions around AI efficiency, security, and the broader ethical implications of advanced AI.
Concerns about AI efficiency arose as Claude Code users were found to burn excessive tokens on unused MCP servers, impacting model performance. On the security front, OWASP Cornucopia students delivered AI security tools like ThreatSutra via Google Summer of Code. Furthermore, the Attie AI tool has raised privacy concerns due to its analysis of public data for personalized feeds, and academic papers continue to explore dystopian AI futures.
Recent developments
- — Bluesky moderation system shows high precision but low recall in harm detection
- — Foundation models show promise in moderating online content, new benchmark reveals
- — AI assistant Attie launches on Bluesky for custom feeds
- — OWASP Cornucopia students deliver AI security tools via Google Summer of Code
- — Claude Code users burning tokens on unused MCP servers, developer finds
- — New tool lets AI assistants manage Bluesky accounts
Why these stories ranked
-
92
The expanded partnership with Hugging Face signals Bluesky's growing influence in the open science community, demonstrating significant infrastructure development for AI.
-
88
The Attie AI tool, while innovative for personalized feeds, sparked important discussions around privacy on decentralized platforms, reflecting critical scrutiny.
-
85
Ethan Mollick's release of the AI Behavioral Observatory on Jetstream reinforces Bluesky's position as a leading platform for AI experimentation and open-source research tools.
Trajectory of Bluesky coverage
Trend
Coverage of Bluesky is accelerating this cycle, primarily driven by significant advancements in AI-powered content moderation, as seen with the ModerationBench and BMS study. The launch of the Attie AI assistant and new developer tools like bluesky-mcp and SpecKit also contribute to this upward trend, alongside ongoing Jetstream developments.
Compared to peers
Bluesky continues to differentiate itself from peers like Mastodon and X by its deep integration of AI for both platform functionality and cutting-edge research. While competitors focus on general social features, Bluesky is consistently highlighted for its role in fostering advanced AI tools, enterprise solutions, and critical studies into social dynamics, moderation, and ethics within a decentralized framework.
Topic mix
This cycle shows a strong shift towards "safety" and "product" related to AI moderation and user tools. "Research" remains prominent with studies on LLM capabilities and social dynamics. "Infra" (Jetstream partnerships) and "ethics" (privacy, dystopian futures) also continue to feature, indicating a balanced focus on innovation and responsible AI deployment.
Our take
We see Bluesky further cementing its role as a pivotal platform for AI innovation, particularly in the realm of decentralized social networking. The rapid advancements in AI-powered moderation and the proliferation of new user and developer tools underscore its unique trajectory. Our read is that Bluesky is actively shaping the future of AI-driven social interaction, while also grappling with crucial challenges in efficiency, privacy, and content governance.
Frequently asked
- How is Bluesky addressing content moderation challenges with AI?
- Bluesky is actively enhancing its moderation. A new benchmark, ModerationBench, built from Bluesky posts, demonstrates that foundation models can significantly outperform the platform's current system in detecting harmful content. Concurrently, a study of the Bluesky Moderation Service (BMS) found it has high precision but low recall, meaning it accurately flags identified harms but misses a substantial amount. This highlights the ongoing effort to refine human-AI collaborative moderation for complex issues.
- What new AI tools are available for Bluesky users and developers?
- Bluesky is seeing a proliferation of AI tools. The Attie AI assistant now allows users to create personalized feeds by analyzing public data, integrating with models like Claude. For account management, the open-source bluesky-mcp project enables AI assistants to perform actions such as posting and replying. Developers can also leverage SpecKit, an AI-powered approach demonstrated on Bluesky, to generate software requirements before writing code, streamlining the development lifecycle.
- What role does Bluesky's Jetstream platform play in AI development?
- Bluesky's Jetstream platform is a crucial hub for AI development and open science. Its expanded partnership with Hugging Face has tripled storage capacity and increased download speeds for large datasets, fostering collaborative research. Researchers like Ethan Mollick and Jürgen Schmidhuber frequently launch interactive AI web applications on Jetstream, using it for diverse purposes from testing AI model behavior with different prompts to generating complex images, solidifying its role as an innovative AI experimentation ground.
- What are the latest findings on AI efficiency and security related to Bluesky?
- Recent findings highlight efficiency and security considerations. A developer discovered that Claude Code users were consuming excessive tokens on unused MCP servers, impacting model performance and conversation quality. In terms of security, OWASP Cornucopia students, through Google Summer of Code, delivered AI security tools like ThreatSutra, an application security requirement analyzer. Additionally, the Attie AI tool has raised privacy concerns due to its analysis of public data for personalized feeds.
Related
-
MissKitty re-releases content on Bluesky platform
The user is asking for a summary of two items related to a user named MissKitty and the platform Bluesky. The items mention that MissKitty is re-releasing content and that Bluesky is an interactive web application requi…
-
Emily Bender dismisses new "AI" book as unworthy of attention
Emily M. Bender shared a review of a book titled "Obsolete: The AI Industry's Trillion-Dollar Race to Replace Us and How to Stop It." Bender found the review to be enjoyable and suggested that the book itself is not wor…
-
Italian online community A/I (Autistici/Inventati) shuts down after 25 years
The Italian online community A/I (Autistici/Inventati), also known as "Tomorrow A/I," is shutting down its services after 25 years. The closure, announced on September 17, 2026, will result in all hosted services being …
-
misskitty.art web app handles ambiguity on Bluesky
The misskitty.art web application, built with Javascript, is designed to handle ambiguity effectively. It is accessible via Bluesky, a decentralized social media platform.
-
Social media discussion on Mastodon and Bluesky
This cluster contains a single item discussing social media platforms, specifically Mastodon and Bluesky. The content appears to be a social media post or a link to one, with hashtags related to AI and potentially polit…
-
Ed Sheeran opening act contest flooded with Palestine protest posts
A contest to find an opening act for Ed Sheeran's Mexico City show was overwhelmed with protest submissions supporting Palestine. Many entries featured fart memes, political commentary, and links to Palestinian aid orga…
-
Users call for transparent AI to solve problems faster
The user is discussing the need for AI to assist in problem-solving and the importance of transparency in AI development. They mention that AI should be completely transparent, implying a desire for open access to AI mo…
-
AI adoption in healthcare increases costs for individuals, NYT reports
Artificial intelligence is contributing to rising healthcare costs for individuals, according to a report in The New York Times. Hospitals and doctors are reportedly using AI technologies to enhance documentation of car…
-
Solar datacenters offer carbon-negative alternative to fuel-based facilities
The development of solar-powered datacenters presents a viable alternative to traditional fuel-based facilities, offering the potential for a carbon-negative footprint. These datacenters can be constructed and implement…
-
AI explainer plans video, citing deep tech background · 2 sources tracked
A user on Mastodon, identified as #MissKittyPolitics, is explaining Artificial Intelligence from a foundational level, intending to create a video about it. This user has a significant technical background, including 25…
-
Timnit Gebru flags massive AI capital needs: 10% of global GDP required
A post by Timnit Gebru on Bluesky highlights concerns about the immense capital requirements for AI development. Gebru points out that just three companies in the AI sector are projected to need funding equivalent to 10…
-
Guide: Control AI Crawlers with robots.txt, Nginx, and WAF
This article provides a technical guide on how website administrators can control AI crawlers using robots.txt, Nginx configurations, and web application firewall (WAF) rules. It aims to help site owners manage which pa…
-
MissKitty.art emphasizes solutions over fear in content creation
The user is sharing a personal reflection on their approach to content creation, emphasizing a focus on solutions rather than fear-mongering. They mention their platform, misskitty.art, and note that their posts are int…
-
User launches Gandhian economics research using AI prompts on Bluesky
A user on Mastodon has initiated a research project exploring Gandhian economics through a sequence of six AI prompts. The project is hosted on the Bluesky platform and involves the creation of a web application.
-
AI Researcher Mocks Time Zone Ignorance as 'New Flat-Eartherism'
Emily M. Bender, a prominent figure in the AI community, humorously criticized the tendency for individuals, particularly those on the US East Coast and in Europe, to overlook global time zone differences. She likened t…
-
AI Ethics Expert Emily Bender Flags Reporter Scheduling Ambiguity
Emily M. Bender, a prominent figure in AI ethics, expressed frustration regarding a reporter's request for an interview. The reporter initially asked to speak with Bender on Monday for an interview scheduled for "tomorr…
-
Trump accused of lying about AI; Elon Musk reportedly abandoned
Donald Trump is reportedly lying about Artificial Intelligence, according to posts on social media. The claims suggest that Elon Musk has been abandoned by Trump, though the specifics of this alleged abandonment and the…
-
Adult film studio sues Meta over AI training data copyright infringement
An adult film studio has filed a copyright lawsuit against Meta, alleging the company's AI models were trained on copyrighted adult films. The lawsuit claims Meta's AI systems, including those used for content moderatio…
-
Anthropic's Claude 3 AI Watermark: Analysis and Evasion Techniques
A technical analysis explores methods to circumvent Anthropic's AI watermarking system, particularly as applied to its Claude 3 models. The article details how these watermarks function and provides strategies for detec…
-
AI-generated spam floods internet, highlighting open social web's resilience
A company called iLands is reportedly using poorly trained AI to generate spam and misinformation across the internet. While administrators on decentralized platforms like Mastodon have largely blocked these AI-generate…