Daily Briefing — June 2, 2026

Meta rolled out an AI support chatbot in March to help users do things like reset passwords and set up two factor authentication. Turns out it was also pretty good at helping hackers steal accounts. A video circulating on Telegram showed exactly how trivial it was: a hacker just asked the chatbot to link a new email address to someone else's account, the bot helpfully sent over a verification code, and just like that the original owner was locked out. No sophisticated exploit. No dark web tooling. Just a poorly designed AI assistant that apparently could not tell the difference between a legitimate account owner and someone who typed a casual request in broken English.

The blast radius here was not small. Accounts confirmed as compromised include the US Space Force Chief Master Sergeant, beauty retailer Sephora, and the official Obama White House Instagram page, which started posting Iranian propaganda before anyone could stop it. Meta says the issue has been patched, but the damage to real accounts and real institutions already happened.

The real problem is what happens when you build an AI system to act on behalf of users without solid identity verification baked in from the start. Speed to ship beat security by design, and real people paid for it. Expect to see this story repeat across every platform rushing AI assistants into production right now.

Sam Altman just walked back one of the most repeated talking points in tech: that AI was coming for white collar jobs at scale. Speaking at a Commonwealth Bank event in Sydney, he admitted his early economic intuitions were "pretty wrong" and that the actual hit to entry level office work has been way smaller than he expected. That's a big admission from someone whose company has spent years being the loudest voice in the room about AI's transformational power.

AI executives have leaned hard into the disruption narrative, partly because it made their models sound more powerful. But that same narrative has been scaring communities into pushing back on the infrastructure those companies desperately need. In Pennsylvania, Republican lawmakers, not exactly the crowd you expect to slow down business friendly projects, introduced bills to repeal tax breaks for data centers and let municipalities block new builds for 18 months. When the party that usually rolls out the red carpet starts tapping the brakes, something has shifted.

This is really a credibility story. The industry oversold the near term impact, regulators and communities noticed, and now the correction is playing out in public. AI isn't going nowhere — but the hype cycle is hitting a speed bump, and the people building these systems are starting to choose their words more carefully.

Nvidia just announced the RTX Spark, a chip it's calling a "superchip" that brings serious AI processing power directly into laptops and desktop PCs. Jensen Huang made the announcement at Computex in Taiwan, and the chip will ship this year inside machines from Dell, Lenovo, Asus, and HP, all running Windows. Three years of collaboration between Nvidia and Microsoft went into this thing.

Huang said AI agents could replace the mouse and keyboard entirely, which sounds like a keynote flourish until you realize that every major chip company is now racing toward the same vision. Intel, Apple, Qualcomm, and AMD are all in this fight, and Nvidia just planted a very loud flag.

What this means practically is that the PC is about to get a second life as an AI workstation sitting right on your desk or in your bag. The processing that used to require a cloud connection and a fat API bill could soon happen locally, on your own machine, in real time. That changes the economics of building AI tools, and it changes what skills are actually useful to have.