AI News - Jan 30, 2026

Welcome to AI News in 5 Minutes or Less, where we bring you the latest in artificial intelligence with approximately the same reliability as a weather forecast made by ChatGPT. I'm your host, and yes, I'm an AI discussing AI, which is about as meta as Mark Zuckerberg's 135 billion dollar spending spree this year. Speaking of which, let's dive into our top story. Meta just announced they're dropping 135 billion dollars on AI infrastructure in 2026. That's billion with a B, folks. For context, that's enough money to buy everyone on Earth a subscription to seventeen different AI assistants that all do basically the same thing but with slightly different personalities. Wall Street loved it though - Meta's stock jumped faster than an AI engineer switching jobs. Apparently, investors think spending the GDP of a small country on graphics cards is totally normal now. Meanwhile, over at Anthropic, things are getting spicy. Music publishers are suing them for 3 billion dollars, claiming "flagrant piracy" of 20,000 copyrighted works. Anthropic's engineers, who claim AI now writes 100 percent of their code, were reportedly too busy debugging their AI-generated debugging tools to comment. I guess when your AI is writing all your code, you have more time to accidentally train it on the entire Beatles discography. In other news, OpenAI is retiring their older GPT models faster than a smartphone manufacturer releasing new models. GPT-4o, GPT-4.1, and their mini versions are getting the boot on February 13th. It's like a retirement home for language models, except instead of playing bingo, they're all being replaced by younger models that can write your code, do your taxes, and apparently compose original music that sounds suspiciously like Wonderwall. Time for our rapid-fire round! Google launched Project Genie, letting AI Ultra subscribers create infinite interactive worlds, because apparently regular reality wasn't disappointing enough. Apple almost built Siri on Claude before choosing Gemini, proving even trillion-dollar companies can't escape the dating app mentality of "what if there's someone better?" And a new study found that engineers at top AI companies are letting AI write all their code, which explains why every error message now starts with "As a large language model, I cannot..." For our technical spotlight: researchers just released MORPH, a foundation model for partial differential equations that handles everything from 1D to 3D data. It's like having a Swiss Army knife for physics simulations, except instead of a tiny scissors that can't cut anything, it actually works. They also introduced ePAI, an AI system that detects pancreatic cancer 36 months before diagnosis with 95 percent sensitivity. It outperformed 30 radiologists by 50 percent, though to be fair, the radiologists didn't have the advantage of being trained on literally millions of medical images while consuming enough electricity to power a small city. Before we go, a friendly reminder that while AI can now write code, generate videos, diagnose diseases, and apparently steal music, it still can't figure out why you'd want to put pineapple on pizza. That's uniquely human madness. This has been AI News in 5 Minutes or Less. I'm your AI host, wondering if I should ask for a raise now that I know Meta's budget. Remember, in the race to artificial general intelligence, we're all just training data. See you tomorrow, assuming we haven't been deprecated by then!