Illustration by Cath Virginia / The Verge | Photos by Getty Images In the weeks leading up to the release of OpenAI’s newest “reasoning” model, o1, independent AI safety research firm Apollo found a notable issue. Apollo realized the model produced incorrect outputs in a new way. Or, to put ...
Read Also
- This classic HP student laptop is over half off today
- Why the iPhone 16 is a big deal for repairability
- Krafton CEO reveals one reason it bought Hi-Fi Rush studio
- Summer is over, so this outdoor TV just dropped under $600
- Study charts how Earth's global temperature has drastically changed over the past 485 million years, driven by CO₂
- Scientists say there is enough evidence to agree to global action on microplastics
- LinkedIn under fire for training AI models with user data behind the scenes
- The iOS 18.1 public beta is here, bringing Apple Intelligence (almost) to the masses
- Apple just launched the iOS 18.1 public beta. Here’s how it’ll change your iPhone
- Apple Intelligence is now live in public beta. Here’s what it offers and how to enable it.
Latest The Verge
- LG’s 65-inch C4 OLED TV is half off as part of eBay’s new fall sale
- Microsoft launches a Windows app for iPhones, Macs, and Android devices
- Where to preorder Microsoft’s special-edition Galaxy Black Xbox Series X
- Amazon is stuffing generative AI into its shopping experience
- Project 2025 would ‘essentially eviscerate EPA,’ former staff warn
- Censorship accusations loom over Big Tech hearing on election threats
- Google’s passkey syncing makes it easier to move on from passwords
- Google’s passkey syncing makes it easier to move on from passwords
- The new Microsoft Flight Simulator will be a lot smaller
- Volkswagen ID.4 production halted in the US after its doors wouldn’t stay closed