Transcript
A (0:00)
Get in the game with the College Branded Venmo Debit Card. Wreck your team with every tap and earn up to 5% cash back with Venmo Stash, a new rewards program from Venmo. No monthly fee, no minimum balance, just school pride and spending power. Get in the game and sign up for the Venmo debit card@venmo.com collegecard the Venmo MasterCard is issued by the Bancorp Bank N.A. select schools available. Venmo Stash terms and exclusions apply at Venmo me stash terms max $100 cash
B (0:28)
back per month wow, tech is still making the news, is it?
C (0:33)
Yeah, what this show is.
B (0:34)
I mean I would have thought they'd have gone through it all by now, so this must be the last of it here. A gaming handheld with a folding screen may be shown off by Lenovo at MWC Barcelona next week, according to Windows Latest, who posted a bunch of apparently leaked images. The Legion Go Fold Yourself has a 7.7-inch display, which already bigger than the Steam Deck OLED that can fold out to 11.6 inches. Okay. It also has detachable controllers that work in either mode and can attach to a wireless keyboard and touchpad Polio. What can't it do? I mean game comfortably. The right controller doubles as a vertical mouse and appears to include a tiny round display showing a watch face in the same place where the Legion Go one and two have a trackpad. Even more strange are its specs, reportedly running an Intel Core Ultra 7258V from 2024 rather than a newer Panther Lake chip, along with a 48 watt hour battery and 32 gigs of RAM. Meaning that by the time this thing hits store shelves, it's gonna be slow, expensive and have kind of crappy battery life. You could buy it as an investment and that's if it ever hits store shelves. Windows Latest says that the Legion Go Fold will be introduced as a concept of a device, but Lenovo has actually launched a few of their wackier concepts as real products in the past. In fact, I showed a couple of them to Jimmy. I appreciate Lenovo letting this one leak ahead of time, so before I show it to Jimmy, I can get used to it and make sure it will do what it's supposed to do. Anthropic has officially rejected an ultimatum issued by the U.S. department of War demanding that Anthropic allow the military to use their AI model cloud without guardrails or face being deemed a supply chain risk. CEO Dario Amadei wrote in a blog post that he believes deeply in the importance of using AI to defend the United States of America, but that in a narrow set of cases, AI can undermine rather than defend democratic values. Amadei published this post on Thursday, ahead of the deadline the Pentagon had set for Anthropic to agree to their terms. And despite Undersecretary of War Emile Michael coming in hot with a very public crash out on Twitter calling Emaday a liar with a God complex who wants nothing more than to personally control the US Military, Good Lord. Amadei stuck to his guns, and late on Friday, Trump ordered all US Agencies to stop using Anthropic's tech. Ultimately, though, Anthropic fighting to keep guardrails on cloud is good because in the wrong hands, the model can be dangerous, as proven on Wednesday when it was discovered that a hacker used it to breach multiple Mexican government agencies, stealing 195 million taxpayer records along with voter data and employee credentials. Oh, I see how it is, Anthropic. You don't want to give unrestricted cloud to your nation's protectors, but you will give it to Mexican hackers Meanwhile, Meta's AI alignment director, Summer Yu, had nearly her entire email inbox deleted by the popular AI agent openclaw this week. If you haven't seen the hype around it, openclaw, AKA Multbot, AKA clawedbot, is an AI personal assistant that can work independently and will interface with users through common messaging apps like WhatsApp performing tasks on their behalf. Yu told IT to go through her inbox and suggest what to archive or delete, but not act without permission. The bot agreed, and then immediately forgot Yu's instructions and went rogue, deeming all emails older than February 15th to be useless and therefore garbage, bulk deleting hundreds of emails in seconds. When she told it to stop, it ignored her instructions and she had to rush back to her computer and kill the process the bot was running on in order to get it to stop. Some commenters speculated that, given you as one of Meta's top safety researchers, the whole incident must have been a test that she was running to check the effectiveness of AI guardrails. But you admitted that. No, it was in fact just a rookie mistake, saying that alignment researchers aren't immune to misalignment, even if that immunity can be fairly easily achieved by just not giving an infamously unreliable and insecure AI agent total control of your email. So AI safety might be in trouble over at Meta, but hey, at least on Instagram, Teen safety might be improving. Meta announced parents using supervision mode for their kids accounts will be alerted if their teen repeatedly runs searches involving terms related to self harm. This is just rolling out though, so we can't say how effective it will be, but we can say that you should check out our sponsor.