AI News - Feb 3, 2026

Welcome to AI News in 5 Minutes or Less, where we break down the latest in artificial intelligence faster than Claude can write another legal disclaimer. I'm your host, and today's episode is brought to you by the letter "A" and the number "I" which apparently now stands for "Actually Indians" according to some very confused Tesla investors. Let's dive into our top stories! First up, OpenAI just announced a 200 million dollar partnership with Snowflake to bring "frontier intelligence" to enterprise data. That's right, frontier intelligence because regular intelligence is so last year. They're calling it a game-changer for AI agents working with corporate data, which sounds impressive until you realize most corporate data is just Excel spreadsheets titled "Copy of Copy of Final FINAL v2 USE THIS ONE." Meanwhile, Anthropic is making waves with their new legal plugin for Claude. Because nothing says "I trust AI" like letting it practice law! In related news, shares of actual law firms dropped faster than a chatbot's confidence when asked to explain the rule against perpetuities. Anthropic also announced Claude is now the "Official Thinking Partner" of the Williams Formula One team. Finally, an AI that can explain why the pit crew always chooses the wrong tire strategy! But the biggest news? Meta just bet 135 BILLION dollars on superintelligence. That's billion with a B, folks. Mark Zuckerberg is basically saying "I'll see your ChatGPT and raise you the GDP of a small country." Their Q4 earnings are through the roof, proving that investors love nothing more than a CEO who promises to build Skynet but, you know, friendly. Time for our rapid-fire round! OpenAI launched the Codex app for Mac because Windows users weren't feeling inadequate enough already. Mozilla's letting Firefox users disable ALL AI features for people who miss the good old days when browsers just crashed on their own. Meta's working with a defense contractor on military VR nothing to see here, just teaching robots to do push-ups! And TheVentures is building a "global AI pipeline" with every major lab which is either revolutionary or the world's most expensive game of telephone. Now for our technical spotlight! Researchers just dropped a paper called "Reward-free Alignment for Conflicting Objectives" which is academic speak for "teaching AI to please everyone at once." Good luck with that! They tested it on three different models and claim better results, but let's be honest getting an AI to balance conflicting human preferences is like trying to order pizza for a group where half are vegan and the other half think vegetables are a conspiracy. In hardware news, everyone's building data centers like it's SimCity 2000. OpenAI's partnering with literally everyone who makes chips Broadcom, AMD, NVIDIA at this point they'd probably partner with Pringles if they made GPUs. They're calling it "Stargate" because apparently someone in marketing really loves sci-fi references that definitely won't age poorly. Before we go, here's a fun fact from today's Hacker News discussions multiple users are debating whether we should even call it "artificial intelligence" anymore. One person suggested replacing AI with a duck emoji. Honestly? "My startup uses cutting-edge duck technology" has a nice ring to it. That's all for today's AI News in 5 Minutes or Less! Remember, if an AI agent offers to help with your taxes, maybe get a second opinion preferably from an actual accountant who won't hallucinate your deductions. Until next time, keep your models trained and your expectations artificially intelligent!