|
🧳 Meta's Muse wants to run your digital errands
|
|
.....so what did Meta actually launch?
|
|
Meta recently released Muse, not to be confused with Muse Spark, a personal AI agent that sends emails, books travel, fills out forms, haggles your bills, and checks out through Link by Stripe. US only for now, on web, iOS, Android, and WhatsApp.
|
Mark Zuckerberg on Meta's Muse AI app and privacy promises
youtu.be
|
|
Would you let Meta's agent read your inbox and hold your card?
|
👀 wait, Meta holding my passwords?
|
|
Meta says it never sees them. Muse runs inside its own secure VM with a built-in browser, a watchdog called Sentinel decides what it can read versus write, and a version Meta itself cannot look into is promised by year end.
|
Take the full tour of Muse, Meta's personal AI agent.
youtu.be
|
|
Yeah, there’s no getting around that reputation, huh? Two FTC settlements and Cambridge Analytica are the elephant in the room. Meta gets your query data unless you opt out, though it insists conversations never touch its ad systems.
|
☠️ An Anthropic researcher quit and said the quiet part out loud
|
|
🤔 why is everyone sharing this guy's post?
|
|
Jacob Coxon, 27, built models at OpenAI and then Anthropic. On September 8 he quit on X, saying both labs are "gambling with our lives," and 90 million people read it in a day.
|
Extended Interview: Ex-Anthropic researcher's AI warning
youtu.be
|
|
Do you REALLY think AI could wipe out humanity?
|
🤔 where does the 10% come from?
|
|
His old boss's boss. Anthropic's Evan Hubinger replied that he puts the odds of AI killing all humans at more than 10% within a decade, and Hinton, Musk, and Dario Amodei have all said similar or worse.
|
📢 so is this real or marketing?
|
|
Depends who you ask. Bernie Sanders seems to think it’s absolutely true. Coxon points at 10,000 OpenAI agents solving a Millennium Prize problem in 88 hours; critics point at an Anthropic IPO a few weeks away.
|
🎯 weekly challenge: hand one chore to an AI this week
|
Muse is US only for now and brand new, so many still can’t try it. This week, fake it: pick one recurring chore and make ChatGPT, Gemini, or Claude do the thinking while you do the clicking.
🗓️ Step 1: Pick the chore Something you do weekly and quietly hate: meal planning plus the grocery list, comparing flights, or checking a phone bill for junk fees.
📸 Step 2: Hand over the raw inputs Photo of the fridge, screenshot of the bill, or the three flight options pasted into ChatGPT, Gemini, or Claude. Ask for the finished output, not advice.
✉️ Step 3: Have it write the message Have it draft the email to your carrier, the text to your partner, or the order note. You paste, you don't compose.
⏱️ Step 4: Time it If it saved you 20 minutes, that is the case for an agent. If not, that is the case against paying $20 for one.
Is a $20 agent that books your flights worth trusting Meta with your inbox, or is the free tier the real trap? And if you believed the 10% number, what would you build with Muse's free tier, and what would you refuse to? We'd love to hear your thoughts!
