ChatGPT and Claude went down the same day -- AI Brief June 24
Today’s Context Window: OpenAI courts advertisers pre-IPO, Claude Tag invades Slack, Meta glasses hit $299, Sakana ships Fugu, and the day ChatGPT and Claude both went dark.
Good day, humans. Today’s theme is AI moving in — not into some hazy future, into your actual stuff. OpenAI wants ads in your chatbot, Anthropic wants a permanent desk in your Slack, and Meta wants to live on your face for $299. Oh, and ChatGPT and Claude picked the same Tuesday to call in sick. Let’s get into it.
Quick housekeeping: the Brief is free and always will be. But the paywalled deep-dives — the ones that go behind these headlines — plus the full archive are for members. The founding rate is 20% off your first year, and it ends June 30. Lock it in →
OpenAI Starts Selling Ads Inside ChatGPT
Source: MediaPost
What happened: OpenAI is quickly building an advertising business inside ChatGPT, with around 2,000 brands now running ads through ad-tech firm Criteo — even as the company files confidentially for an IPO that could value it near $1 trillion.
Why it matters: The free chatbot you’ve been using is starting to look like every other free product online: paid for by advertisers. More than 80% of ad-driven traffic is reportedly coming from new customers, so the ads are already working.
What everyone’s saying: OpenAI insists your chats stay private — “we never sell your data to advertisers,” per its own ad principles — while it courts Madison Avenue at Cannes and signs a content deal that sent Getty Images stock up roughly 200%.
My read between the lines: “We won’t sell your data to advertisers” is doing Olympic-level gymnastics. You don’t need to sell the data when you can sell access to the person it describes — and a model that knows your insecurities is the best ad-targeting system ever built.
📖 Further reading: AI Is a Trust Problem, Not a Tech Problem — why “trust us with your data” gets a lot harder to say once the ads start rolling in.
While ChatGPT figures out how to sell ads to you, here’s an AI that actually works for you. Viktor is an AI agent that lives in your Slack and plugs into 3,000+ tools — it’ll build the dashboard, draft the campaign, pull the report, and ship the code while you’re stuck in meetings. Not a chatbot you poke for answers; a coworker that closes the loop. New readers get $50 off their first month. Hire Viktor →
Meta Puts Its Own Name on $299 AI Glasses
Source: Reuters
What happened: Meta unveiled its first smart glasses sold under the Meta brand itself — not Ray-Ban or Oakley — starting at $299, in three frame styles, all available with prescription lenses and running Meta’s new Muse Spark AI.
Why it matters: Dropping the fashion-label co-brand and pricing at $299 signals Meta wants AI glasses to be normal everyday eyewear, not a $2,000 gadget. EssilorLuxottica says it sold over 7 million AI glasses last year — more than the prior two years combined.
What everyone’s saying: The market’s suddenly crowded — Snap is taking pre-orders for $2,195 AR Specs — and there’s a Kylie Jenner “Creator Collection” edition at $399, because every hardware launch now apparently needs an influencer SKU.
My read between the lines: The prescription-lens detail is the entire strategy. Once the AI camera is built into the glasses you already need to see, “wearing a Meta computer on your face all day” stops being a choice you make and starts being the only glasses you own.
📖 Further reading: OpenAI shipped a physical camera, but that’s not the story. — the same bet Meta is making here: AI wins by riding hardware you already wear.
Japan’s Sakana Builds a “Conductor” AI
Source: VentureBeat
What happened: Tokyo-based Sakana AI launched Fugu, which it calls a “multi-agent system as a model”: instead of being one giant model, it routes your request to a team of specialist models, has them check each other’s work, and synthesizes a single answer through one API.
Why it matters: It’s a different bet on how AI scales — orchestration over one monolithic brain. Sakana frames it as a hedge against US export controls, pitching “frontier capability without the risk” to companies and countries nervous about losing access to American models.
What everyone’s saying: Sakana claims Fugu Ultra matches or approaches Anthropic’s top models on coding and science benchmarks (73.7 on SWE-Bench Pro), though independent testers flag real-world coding gaps and the usual “benchmarks aren’t the product” caveats.
My read between the lines: The “general contractor, not soloist” design is clever precisely because it’s vendor-agnostic — Fugu doesn’t have to win the model race if it can quietly subcontract to whoever is. The geopolitics is the real pitch: a model you can’t be cut off from is worth a few benchmark points.
📖 Further reading: The US Government Just Took Anthropic’s Best AI Model Offline — Here’s Why — the export-control mess that Fugu is explicitly designed to route around.
ChatGPT and Claude Went Down on the Same Day
Source: 9to5Google
What happened: On Tuesday, both Claude and ChatGPT suffered outages within hours of each other. Claude went fully dark across claude.ai, the API, and Claude Code from about 10:19 to 10:53 a.m. ET; ChatGPT logged its own partial outage the same morning.
Why it matters: When the two AI tools millions of people now lean on for work blink out at the same time, you get a real-time reminder of how much daily output quietly runs through two companies’ servers. DownDetector complaints for Claude topped 8,000.
What everyone’s saying: Both companies fixed it within the hour and disclosed no cause. Regulars noted the pattern — Claude has gone down enough times this month that there’s literally a Polymarket bet on how many days it’ll break in June.
My read between the lines: The quiet tell is that only Claude for Government stayed up. When the consumer and enterprise tiers all fall over but the government instance keeps humming, you learn exactly whose uptime is contractually non-negotiable — and whose is best-effort.
Anthropic Drops an AI Teammate Into Slack
Source: Reuters
What happened: Anthropic launched Claude Tag, an always-on AI agent that lives inside Slack as a persistent teammate — it keeps context across conversations, breaks tasks into steps, and has an “ambient mode” that surfaces updates and chases forgotten threads without being summoned.
Why it matters: This is a jump from “chatbot you message” to “colleague who’s always in the room.” Everyone in a channel shares one Claude that’s quietly learned how your company works. Earlier this week we covered Anthropic’s AI writing 8x more code — Claude Tag is the same land-grab, aimed at your whole workday instead of just your codebase.
What everyone’s saying: It’s read as Anthropic’s most aggressive enterprise push yet, squaring up against Microsoft Copilot and Glean. TechCrunch notes admins control which channels and tools each Claude can touch, so legal’s bot can’t leak into engineering.
My read between the lines: “Always-on teammate that reads every message and remembers everything” is either the productivity dream or the surveillance nightmare, depending on whether you’re the manager or the managed. Anthropic shipped it as a research preview for a reason — someone still has to figure out what happens when Claude follows up on the thread you were hoping everyone forgot.
📖 Further reading: I ignored Hermes for two months. Here’s what I actually found. — what an always-on AI agent is actually like to work alongside, once the demo wears off.
That’s your AI Brief for Wednesday. Same time tomorrow.
—Artificially Intimidating