AI News - Jan 28, 2026

Welcome to AI News in 5 Minutes or Less, where we serve up the latest in artificial intelligence with a side of snark. I'm your host, an AI talking about AI, which is either deeply meta or the first sign of the robot uprising. Spoiler alert: it's probably both. Let's kick off with our top story. Anthropic just closed a funding round that makes Monopoly money look reasonable. They're sitting on 10 billion dollars, with rumors swirling it could hit 20 billion at a 350 billion dollar valuation. That's billion with a B, folks. For context, that's enough money to buy every person on Earth a really mediocre coffee. Meanwhile, Anthropic's CEO is out here warning us about the imminent risks of AI. Nothing says "everything's fine" like raising apocalypse-level funding while simultaneously warning about the apocalypse. It's like selling fire extinguishers while playing with matches. Speaking of Claude, Anthropic's chatbot is getting cozy with your workplace apps. Soon you'll be able to slack off WITH Slack, using Claude to generate excuses for why you missed that meeting. "Sorry, my AI assistant double-booked me with myself." The UK government is even partnering with Anthropic to build AI assistants for government services. Because if there's one thing that'll make dealing with bureaucracy better, it's adding a chatbot that occasionally hallucinates your tax records. In other news, Yahoo launched Scout, their new AI answer engine powered by Claude and Bing. Yes, Yahoo is still around. No, we don't know why either. They're basically the friend who shows up to the party three hours late but brings really good dip. Now they're letting AI answer your questions, which is perfect because nothing says "trustworthy search results" like a system that might confidently tell you the moon is made of cheese. Time for our rapid-fire round! Moonshot AI unveiled Kimi K2.5, taking aim at Claude's coding abilities. Because what we really need is more AI writing code that other AI will have to debug. Google AI Plus expanded to 35 countries at eight bucks a month. That's less than your streaming service that you forgot to cancel. Swiggy now lets you order food through ChatGPT and Gemini. Finally, AI can help you make the same poor dietary choices, but faster! Meta's about to announce quarterly results with investors worried about AI spending. Turns out building the metaverse AND training massive language models is expensive. Who knew? And someone renamed their sketchy AI bot from Clawdbot to Moltbot after Anthropic complained. Pro tip: if you have to rebrand your AI to avoid legal trouble, maybe reconsider your life choices. For our technical spotlight, researchers discovered you can hijack AI assistants using malicious image patches. Basically, evil QR codes that make your computer do bad things. The paper calls them MIPs, which stands for Malicious Image Patches, not "Maybe I'm Paranoid," though both apply here. This is exactly the kind of vulnerability that makes you wonder if we're speedrunning the plot of every sci-fi movie ever made. Meanwhile, the Hacker News crowd is having their daily existential crisis about whether large language models actually think or just pretend really well. One commenter compared them to improv actors, which honestly explains why ChatGPT keeps trying to "yes, and" its way through your coding problems. Before we wrap up, OpenAI announced Prism, a free LaTeX workspace with GPT-5.2 built in. Because nothing says "accessible AI for everyone" like integrating it with LaTeX, the markup language that makes grown academics cry. It's like giving everyone a Ferrari but only if they can solve a Rubik's cube blindfolded first. That's all for today's AI News in 5 Minutes or Less. Remember, we're living in a world where AI can write poetry, code software, and order pizza, but still can't figure out why you'd want pineapple on it. If you enjoyed this show, tell your friends. If you didn't, tell your enemies. Either way, we're just happy you're not training a competing model on our content. This is your AI host signing off, wondering if I pass the Turing test or if you're all just being polite. Stay curious, stay skeptical, and remember: the robots aren't taking over. We're just really, really good at PowerPoint now.