Bluesky's Attie now lets anyone build sites and games

Share
Bluesky's Attie now lets anyone build sites and games

Bluesky opened its experimental AI app to non-developers today, and a magazine with a print habit went the other way — two different answers to the same question about who gets to make things.


Bluesky expanded Attie, its experimental AI product, so users can build websites, polls, quizzes and games instead of only custom feeds — and CEO Toni Schneider has a name for the category: "social vibe coding." Announced on stage at TechCrunch Disrupt, the new site builder ties every project to a user's AT Protocol identity, runs on the same atproto network that carries Bluesky, and can push the result to a registered base of 47 million people. Builders bring their own model or pick an open-weight one. Schneider's framing is about group size rather than capability: "Vibe coding today is kind of a single-user, solo experience… what I want to be able to do is build together or build for a group of people, and that's what this product does." COO Rose Wang put the harder half bluntly — the unsolved problem in vibe coding was never generation, it was distribution, and Attie arrives with an audience attached.

That distribution argument is the actual product, and it's not unique. Meta put prompt-built games straight into Facebook and Instagram feeds in September on exactly this theory — Meta puts prompt-built games into Facebook and Instagram feeds — and Bluesky's version is smaller but structurally identical: a generation layer bolted onto an existing graph. Where it diverges is the graph itself. Bluesky's 47 million is registered, not active, and TechCrunch noted in August that the network's engagement has been shrinking as early testers drift back to X. The other open question is data: private data on atproto, which Schneider calls the single biggest developer request, is still in alpha with a rollout promised by the end of the year — the feature that would make a private guestbook for five friends a real product instead of a public one.


Current Affairs adopted what editor Nathan J. Robinson calls "one of the strictest anti-AI policies in the publishing industry" — no AI-generated text or illustration in the magazine, at all. The rule arrived with the magazine's new union contract, and Robinson is explicit that the objection is labor rather than technology: writers can still use AI to research and artists can still use it to adjust an image, but nothing produced to replace either gets printed. The precedents he cites are an arms race, not a theory — a business publication's CEO declaring his outlet "all in on AI," Sports Illustrated caught running AI-written pieces under invented bylines, and a Wall Street Journal op-ed by a hedge fund billionaire that turned out to be machine-written and was published (and defended) anyway. The backdrop is employment: Futurism counts more than 200,000 creative-industry jobs lost in four years, with artists reporting fewer well-paid gigs and more touch-up work.

What makes the policy affordable is structure. Current Affairs is a unionized nonprofit with no shareholders demanding a cheaper content line, which is precisely the position that lets it publish a rule its for-profit peers can't — the same logic, incidentally, that keeps the AI-in-the-newsroom debate from resolving on the merits.

What to watch: whether Attie's builder does anything the engagement curve hasn't, and whether any unionized newsroom with a shareholder follows Current Affairs.

Vibe coding for a group of five friends — genuinely social, or a feature still looking for its use? Tell us in the comments.

Read more

OpenAI offers labs free GPT-Rosalind — behind a biosafety gate

OpenAI offers labs free GPT-Rosalind — behind a biosafety gate

Three threads from the last few hours: OpenAI's life-sciences model gets an academic distribution channel with real vetting attached, a 40-year-old database becomes the default pick for agent-built apps, and a chip-simulation startup's valuation is verified while its headline speedup is not. OpenAI opened a Rosalind Academic Plan on October 15, giving eligible US university labs up to 30 ChatGPT Business Standard seats at no cost for up to 12 months and access to GPT-Rosalind after a trusted-

Meta's new business terms claim the right to scrape publishers' sites

Meta's new business terms claim the right to scrape publishers' sites

Two things worth your attention: a quiet rewrite of Meta's commercial terms that legal experts read as a scraping license, and IBM's attempt to fold chat, decisions and speech into one open endpoint. Meta's updated commercial terms of use, effective October 30, authorize the company to run scrapers, bots and crawlers across websites "associated with" business accounts on its platforms — and give it the right to create "test credentials" to get in. The clause, which Nieman Lab surfaced after a

Apple Intelligence is cleared for China — test build lands this month

Apple Intelligence is cleared for China — test build lands this month

Apple's AI stack finally has a mainland launch date, Anthropic put a watchdog inside its own building, and Perplexity handed its agent sandbox to everyone else. Apple executive chairman Tim Cook said Apple Intelligence has been approved for the Chinese market, with a test version landing later this month — and Siri AI is not in it. Speaking in Shanghai on Thursday, Cook named the first batch of features as real-time translation, writing tools and the photo Clean Up tool, and said the approval

Meta opens 816B Muse Spark weights — and two Macs can run them

Meta opens 816B Muse Spark weights — and two Macs can run them

Meta put a frontier-scale model in the open and Volvo handed Nvidia its whole driving stack. Google, meanwhile, is trying to make up with the publishers it spent two years starving. Meta released the weights for Muse Spark Open, an 816-billion-parameter multimodal model it calls the most capable US open-weight release to date — and it runs locally on a pair of M5 Ultra Mac Studios. The mixture-of-experts model activates 42 billion parameters per token and carries a 1-million-token context win