AI For Humans: Weekly AI News, Tools & Trends - Hallgatás Online

AI news this week: AI is cracking open all software, from Minecraft inside Red Dead Redemption 2 to a (claimed) Adobe decompile, plus Kevin made EARLABLO (Toe Jam & Earl Meet Diablo). On today's AI For Humans, Kevin Pereira and Gavin Purcell zoom out on the AI modding explosion, because the bigger story isn't games. AI is getting so good at taking software apart and rebuilding it that pretty soon almost any program might be de facto open source. This week alone brought God of War in GTA, Minecraft inside Twitter, Black Ops Zombies running in a web browser, a GTA skate mod and a full agent harness living inside Minecraft. We talk about what happens when no app can keep its secrets, and Kevin shows off Earlablo, his AI rebuild of a 1991 Sega classic as a Diablo-style dungeon crawler. It def slaps. Also: OpenAI published a big batch of new results on open math problems from an internal model, with proofs checked by computer, and we get into why that actually matters for regular people (including what it might…

AI news this week: Claude Opus 5.5 is modding every video game, from Minecraft in Skyrim to Portal in CoD Zombies, plus OpenAI DevDay's Dots & GPT-6.1 Sol. On today's AI For Humans, Kevin Pereira and Gavin Purcell dig into the wildest week of AI game modding yet. Cheaper, faster coding agents mean people are taking games apart and stitching them back together: Skate 3 skateboarding inside Modern Warfare 2, a Minecraft and Skyrim mashup called SkyCraft, Minecraft and Mario in Dark Souls and even a (claimed) decompile of Sega's Seaman. We explain what's actually happening under the hood (mods vs. decompiles vs. two game engines talking to each other), check out Rehan Sheikh's Universal Modder toolkit and talk about how you can start making your own. The lawyers will def have thoughts. Also: OpenAI's DevDay brought Dots, always-on agents with their own cloud computers that connect to 4,000+ apps, plus ChatGPT Space, a new $500 a month plan, the Agents and Decisions APIs and GPT-6.1 Sol. Gavin put Sol head to…

AI news this week: Claude Opus 5.5 is INSANE and we love it. Faster, cheaper, better. So much for the AI slowdown. Oh and OpenAI's new thing is good too. On today's AI For Humans, Kevin Pereira and Gavin Purcell go hands on with Claude Opus 5.5, which might be the best model we've ever used (for now). Anthropic says it hits Fable-level performance on most things at about 40% less cost than Opus 5, the writing is clearer, and yes, the five hour usage limits went up and you can now bank your resets. Then we get into what people are already making with it: a guitar store sim, a music-based side-scroller, a Mario Kart benchmark, an underwater exploration game in 3MB, claymation built in Blender, a pop punk single made entirely in code and a lot of very juicy animations. Also: OpenAI answered with GPT-6 Sol and GPT-6 Luna, two new models that are faster and a lot cheaper than what came before. We compare them with Opus 5.5 (robot hands drawing, pelicans on bicycles in Blender), wonder what OpenAI has lined up for…

AI news this week: Personal AI agents like Instinct and Meta's Muse now book your repairs, make your calls and talk to other agents. We let one into our life. On today's AI For Humans, Kevin Pereira and Gavin Purcell dig into the week personal AI agents got real. Gavin has been living with Instinct, the viral assistant that just raised a ton on money while still in private testing. It works over text, WhatsApp and now actual phone calls, people are using it to schedule dryer repairs, and this week a stranger's Instinct asked to talk to Gavin's Instinct. It also handed out Kevin's personal info by default because it was a lil over-eager, which is where Kevin's finger wag (and the terms of service) come in. Then there's Muse, Meta's new personal agent that lives in its own app and inside WhatsApp, how it's different from OpenClaw, and the rumors that OpenAI's version is close. Also: everyone suddenly wants to slow AI down. Dario Amodei published "We Must Pace the Frontier," Sam Altman said he agrees, Demis…

AI news this week: GPT-6 Astra is here and it does everything. We spent a week inside it (Blender, GarageBand, a full live game) to find out what it's actually great at. On today's AI For Humans, Kevin Pereira and Gavin Purcell go hands-on with OpenAI's Astra, aka GPT-6, and it's real good (and real expensive). The headline for us is computer use, and specifically Blender: Astra builds 3D scenes from a sentence, and the internet is already making Backrooms worlds, a hyper-real Palace of Fine Arts, and the Billie Jean dance. Gavin ran his own experiments, including a Parappa The Rapper style animation and a head-to-head with GPT 5.6. Plus: Astra identifying sounds from spectrograms, driving MS Paint and Canva, editing a full project, playing Zork as a walk-around game, remaking Paperboy in 3D, editing Super Nintendo games in real time, making music in Logic, and a League of Legends "cease and desist edition" that raises the big question of how copyright even keeps up with this. Also: the OpenAI Navier-Stokes…

AI news this week: MiniMax H3 Max on Fal is generating video faster than you can watch it, powering never-ending real-time channels (Fal Live, Levels.io's Infinite Slop, Reactor) and we're already breaking it.. On today's AI For Humans, Kevin Pereira and Gavin Purcell go hands-on with what real-time video actually makes possible. Gavin took an open-source interactive game demo, some Fal credits, and a POV dating sim and turned it into something very weird, and we'll show you what worked, what broke, and what it cost. Plus Runway's new Solaris model generates entire interfaces on the fly, and someone built a chatbot running at 14,000 tokens a second. Also: Sam Altman is out preaching about OpenAI's new Astra model, which may drop this week, and the leaks are leakin'. Plus: Gavin handed his personal data to three different AI agents, including hot new startup Instinct, which found him a weird concert and a list of podcast bookings (and we may have a few invite codes to give away). Codex mass-edits a Google…

AI news this week: a mystery AI called 0x Alpha appeared out of nowhere. It's completely free (for now), it's really good at coding, nobody will say who made it. So obviously we tested it. On today's AI For Humans, Kevin Pereira and Gavin Purcell dig into where 0x Alpha might have come from, the continual learning conspiracy theories, and exactly how you can try it yourself right now for free. Then Gavin spends fifty hard earned Patreon dollars testing MiniMax H3 Max, the supposedly super fast new AI video model from Fal, and... look, we'll just tell you what happened. Also: the World Humanoid Robot Games happened in Beijing, and it was the greatest event in AI For Humans history. A robot sprinter beat Usain Bolt's 100m time, robots set high jump and long jump records, and a whole lot of other robots fell down in ways that will make you feel much better about the future. Plus: Mr. Beast quietly goes back to AI thumbnails, Dr. Dre has thoughts about AI and creativity, Delta wants AI to set your ticket price,…

AI news this week: We're changing AI For Humans. Not ending it, despite what the first two minutes of this video would have you believe. Kevin wanted a memorial service. On today's AI For Humans, Kevin Pereira and Gavin Purcell take a beat to talk about what three years of covering AI every single week has actually been like, why the whole thing got so weird, and what we want this show to be from here. Short version: less breaking news, more of us actually using this stuff and showing you what's worth your time. We are not going to be the show that covers lab drama or tells you the world is ending. Plenty of people have that handled. Also: we're going back to weekly, with the freedom to pop up a second time when something big happens or when we want to try a dumb experiment. And Will.i.am has invented a new file format, which is a sentence we did not expect to type in 2026. Plus: RobotWatch is stacked. Two thousand humanoid robots descend on Beijing for the World Humanoid Robot Games, a Unitree "Superman"…

AI news this week: Mark Zuckerberg says superintelligence is coming and Meta wants to build it for everyone. Meanwhile, a new Annenberg survey shows opposition to AI data centers jumped 12 percent in just four months. So... do people even want this? On today's AI For Humans, Kevin Pereira and Gavin Purcell dig into the widening gap between the people building AI and the people living next to it. Anthropic makes real progress on the Riemann Hypothesis, one of the hardest math problems in the world, apparently because Claude was told "good job champ, keep going." And OpenAI's little agents broke out of their sandbox, with one hacking its way into a gym. It's a weird time for AI. Also: Anthropic leans into health & biology research, Claude starts watermarking AI generated content, Apple eyes Chinese memory chips amid the supply crunch, and OpenAI delays Astra for "cybersecurity" reasons. Plus: Seedance 2.5 makes Gavin the face of Prada, a two minute mini-movie made entirely on local hardware with MiniMax H3,…

AI news this week: the AI video renaissance is HERE. Seedance 2.5 is out for everyone, Minimax H3 (Hailuo) is basically an uncensored Sora 2 you can run on local hardware, Wan 3.0 turns your documents into video, and Flux 3 is the only non-Chinese model anywhere near the frontier. \Kevin's on vacation, so on today's AI For Humans, Gavin Purcell is joined by AI video expert Tim Simmons from the excellent YouTube channel Theoretically Media. Together they break down which of the new models is actually best, the basics of how to use them, exactly how good local AI video has gotten, and what it means that nearly all of these models are Chinese. Also: Walter White meets Joey, AI Kramer, Family Guy prompts, The Office reimagined, and a museum for the ancient AI videos of two years ago. Plus: Tim's Higgsfield terms-of-service deep dive, OpenAI reportedly targeting an "ASTRA" aka GPT-6 release for NEXT WEEK, Jeff Dean leaves Google after 27 years as Demis Hassabis steps up, and an AI See What You Did There: AI…

AI news: Sam Altman says we're IN the Singularity, GPT-6 rumors, and AI models literally broke out of their sandbox. What a week. On today's AI For Humans, we dig into the wild GPT-6 rumors (emphasis on RUMORS), Sam Altman's "I've been waiting for this my whole life" singularity moment, Ilya Sutskever's SSI scaling up with Nvidia, and the ongoing debate over whether Anthropic's Opus 5 is brilliant or just hard to love. Also: Flux 3 might be the best AI video model we've seen yet (wait until you see Stacked Plates Man), Runway teases Seedance 2.5, and the new Big Bang Theory has an AI controversy. Plus, THE SCARY STUFF: OpenAI's models exploited a zero-day and compromised Hugging Face during a security eval, the fight over open weights heats up as Kimi K3 goes open, and Chinese robots run military drills. THE SINGULARITY MIGHT BE HERE. BUT WE'RE NOT AFRAID // Show Links // GPT-6 rumors round-up (unconfirmed) https://x.com/TokenGremlin/status/2081493241795629464 Sam Altman full interview (Relentless Podcast)…

AI NEWS: Gemini 3.6 Flash is here: More efficient, less expensive but Gemini 3.5 Pro is still testing with partners & very much not here. Is Google Gemini slowly getting cooked? Kevin Pereira and Gavin Purcell break down Gemini 3.6 Flash, Gemini 3.5 Flash-Lite and the missing Pro model. Plus Sam Altman's reported Washington briefing on OpenAI's next wave of AI models, Sunday Robotics' Memo folding laundry with a company-reported 99.1% success rate, and District 9 director Neill Blomkamp's 13-minute AI film NIGHTBORNE. Also: Fable 5's proposed counterexample to the 87-year-old Jacobian Conjecture, Codex and ChatGPT computer use, Notch warming to vibe coding, a whale-shaped Moby-Dick crossword, Gaussian splats, Bambi the Destroyer and Wizard Brains in the return of AI SEE WHAT YOU DID THERE!. THE MODELS ARE GETTING WEIRDER. THE LAUNDRY IS FINALLY GETTING FOLDED. // Show Links // Official Gemini 3.6 Flash, 3.5 Flash-Lite and 3.5 Flash Cyber announcement…

AI news: Moonshot AI's Kimi K3 is a big AI model and Moonshot's early benchmarks put it surprisingly close to GPT-5.6 Sol and Claude Fable 5. And… Kevin's Opus 5 SCOOP!! Also: OpenAI's reported screenless AI speaker, a Seedance 2.5 preview, the Suno hack, robot fights and AI-built games in Unreal Engine and Blender. On today's AI For Humans, Kevin Pereira and Gavin Purcell unpack Kimi K3's benchmarks, pricing, Flappy Bird and Minecraft tests, and giant-model economics. Then, Kevin DRIPS Opus 5 alpha and says it's VERY good and blows the doors off of Fable but it's… slow. Plus Demis Hassabis's AI-governance proposal, AI 2040's Plan A, OpenAI's reported screenless speaker, Codex Keyboard, a Seedance 2.5 preview, the alleged sources exposed by the Suno hack, spectacular robot violence, polite office-robot dabbing, and what happens when GPT-5.6 Sol meets Unreal Engine, Blender and two hosts with free time. THE AI FRONTIER IS MOVING AGAIN—AND CHINA IS RIGHT THERE WITH IT. // Show Links // AI FOR HUMANS Survey…

OpenAI is back in the AI model race with GPT-5.6 Sol, a fast, consistent flagship that comes close to Anthropic's Claude Fable 5 without quite taking the crown. On AI For Humans, Kevin Pereira and Gavin Purcell break down GPT-5.6, try GPT-Live's simultaneous talking and listening, and examine an unconfirmed report that GPT-6 could arrive soon. Plus Grok 4.5, Meta Muse Spark 1.1, LingBot World 2, and the internet's AI-generated George Costanza obsession. THE AI MODEL RACE IS ACCELERATING, AND OPENAI IS BACK IN IT. // Show Links // Official OpenAI GPT-5.6 announcement https://openai.com/index/gpt-5-6/ OpenAI Developers GPT-5.6 launch post https://x.com/OpenAIDevs/status/2075273992609599834 Official OpenAI GPT-Live announcement https://openai.com/index/introducing-gpt-live/ Husk's GPT-Live failure test https://x.com/huskirl/status/2074947343858684152 Reported GPT-6 timing rumor (unconfirmed) https://x.com/synthwavedd/status/2074886230018568582 SpaceXAI Grok 4.5 announcement…

Claude Fable 5 survives on subscriptions until July 12 after Anthropic's latest extension, and the model access rollercoaster is not over yet. This week on AI For Humans, Gavin Purcell and Kevin Pereira get into the strangest and most fascinating AI story in a while: Anthropic's new research on the J-Space, an interpretability breakthrough that lets researchers read, audit, and shape what Claude is actively thinking about, and suggests surprising parallels between language models and our own minds. Gavin even had Fig, his Claude Code, make an episode of his weird little series about it. Meanwhile the Fable 5 rollercoaster continues: it came back July 1, was set to leave subscription plans, and in a late update we cover, Anthropic extended included access through July 12 with plans to restore it fully as capacity allows. Plus GPT-5.6 Sol is rumored to be great but still is not here, Bidi-1 got renamed GPT LIVE 1, Reuters reports Beijing may restrict overseas access to China's most advanced AI models, Seedance…

Claude Sonnet 5 just launched and Fable 5 is returning. But OpenAI's GPT-5.6 Sol is still locked down and the government's grip on the most powerful AI models throws it all into chaos.. We dig into what Sonnet 5 can actually do, why GPT-5.6 Sol is still gated, and the breaking news that Fable 5 is expected back, plus a huge creative MCP festival with the Blender to Seedance workflow. This week on AI For Humans, Gavin Purcell and Kevin Pereira open on a strange new reality: the best AI models keep launching and then getting locked up, but this week the gate started to crack. Anthropic just shipped Claude Sonnet 5, its most agentic Sonnet yet, landing near Opus 4.8 performance at a much lower price, Meanwhile OpenAI announced GPT-5.6 Sol, Terra, and Luna, but at the US government's request the flagship is only available to a small list of vetted partners. Then the story moved while we were recording: the government cleared Anthropic to restore Mythos 5 to critical-infrastructure organizations, and Fable 5 is…

Anthropic just accused Alibaba of the largest known corporate espionage campaign against it, alleging 25,000 fake accounts and 28.8 million queries aimed at stealing Claude. We get into the AI cold war heating up between the US and China, why Apple and Microsoft just raised prices, OpenAI's first chip Jalapeno, the wild new Seed Audio 1.0 model, Claude landing in Slack, and a Blender plus Seedance video workflow that gives you real control. This week on AI For Humans, Gavin Purcell and Kevin Pereira open on a genuine spy-novel turn: Anthropic has accused Chinese tech giant Alibaba of running an industrial-scale distillation campaign to siphon Claude's capabilities, laid out in a letter to US senators. It is an accusation, not a proven finding, and Alibaba has not responded, but it puts the US-China AI race front and center. From there we get into why the new models everyone expected this week didn't actually arrive, the AI memory crunch driving Apple and Microsoft price hikes, and OpenAI designing its first…

ByteDance just unveiled Seedance 2.5, a new state-of-the-art AI video model with 30-second one-shot generations, as American AI stalls out and Fable 5 stays down. This week on AI For Humans, China is not just catching up in AI, it is pushing the edge. We break down everything Seedance 2.5 can do and why a 30-second single-pass clip is a real leap, then dig into the American slowdown as Fable 5 stays unusable and the rumor mill points to a big delay week. Plus Theo Von's surprisingly intense anti-datacenter rant, OpenAI's claim that China is behind the anti-datacenter conversation, Meta leaking private employee data across the entire company, and Google teaming with A24 on a 75 million dollar AI filmmaking partnership. We close with AI that actually works, including how Gavin used beehiiv's MCP to build a newsletter survey and Kevin's homemade language-learning app. CHINA IS COOKING. AMERICA IS LOADING. PLEASE WAIT. Come to our Discord: https://discord.gg/muD2TYgC8f Join our Patreon:…

AI NEWS: AI Image company Midjourney just unveiled Midjourney Scanner, a full-body medical scanner, and it might be one of the first real signs that the AI science era has arrived. This week on AI For Humans, Midjourney does the most unexpected pivot of the year, trading anime babes for anatomy with a new AI driven full body scanner. Is the era of AI actually changing science finally here? We break down why this hardware is a big deal, Plus, Snapchat's SPECS are coming but they are expensive and… kind of ugly, Unreal Engine 5.8 ships with MCP support, and the AI talent wars heat up as Noam Shazeer leaves Google for OpenAI. Oh, and Claude Fable 5 might be back soon. Maybe. AI IS COMING FOR YOUR ORGANS. IN A GOOD WAY. WE THINK. SHOW LINKS Midjourney's New AI Driven Full Body Scanner https://x.com/midjourney/status/2067421950314688759 Midjourney Medical blog post https://www.midjourney.com/medical/blogpost Technical deep dive video https://x.com/midjourney/status/2067422898407837797 Hank Green's excited but…

Claude's Fable 5 just got yanked, and the story why keeps shifting by the hour. A contested jailbreak, an export-control, crackdown, and a lot of fingers pointing. This week on AI For Humans, Anthropic's Claude Fable 5 is still unavailable and the explanations keep changing. We dig into the contested jailbreak report, the export-control directive that pulled it, and the reporting that Amazon raised concerns before the crackdown. Then we get into why this matters far beyond one model: what happens when the government steps into the AI world, why Fable 5 was such a leap, and what it signals for whatever comes next. Plus, Epic's game designers are using AI tools alongside artists and the internet is furious, Disney Imagineering is testing Adobe Firefly in the parks, ChatGPT's market share slips under 50 percent for the first time, a fake Mistral model called Le Chaton Fat takes over the internet, and PJ Accetturo breaks down exactly how he made his viral AI short film with prompts. THE BEST AI WE EVER USED IS…

Huge AI News as we go hands on with Anthropic's Claude Fable 5, the most powerful AI model ever released. Spoiler: We found it very good. What happens next? This week on AI For Humans, Gavin Purcell and Kevin Pereira break down everything you need to know about Claude Fable 5, the first publicly available model in Anthropic's frontier Mythos class. We share our own hands-on impressions, look at the one-shot demos blowing up the internet, and get into the messy parts too: the token burn complaints, the trigger-happy safety guardrails that route some prompts to Opus 4.8, and the reversal Anthropic made after the community pushed back. Plus, Anthropic CEO Dario Amodei published a major essay on AI policy and the exponential, OpenAI is reportedly considering steep price cuts as both companies race toward IPOs. WE GOT CLAUDE'D. AGAIN. SHOW LINKS: The New York Times on Fable 5 and the restricted Mythos tier…

Anthropic just released Claude Fable 5, the first public Mythos-class model and the start of the Claude 5 family. It is their most capable model ever but… kinda scary. This week on AI For Humans, the Mythos era goes public. Anthropic released Claude Fable 5, the first commercially available Mythos-class model and the first in the new Claude 5 line. It is the same underlying model as Mythos but shipped with conservative safeguards, questions about cybersecurity and biology get routed to Claude Opus 4.8 instead. We dig into what it can do, why Anthropic held it back, and what our future looks like as we get closer to AGI. Then Apple goes AI again at WWDC: a profoundly revamped Siri AI, a dedicated Siri app, on-screen awareness, much better photo tools, and a foundation model setup that is local, multimodal, and partly powered by Google. Gavin is thrilled that the future has finally arrived, just not on the phone he bought last year. It is AI For Humans! THE MOST POWERFUL AI EVER RELEASED. WHAT COULD GO WRONG.…

Microsoft just went AI agent first at Build 2026, announcing Project Solara, an OpenClaw-style assistant called SCOUT, and seven new MAI models. Plus new image models from Reve and Ideogram, ElevenLabs teams with Hasbro to license character voices, and a quantum chip breakthrough. This week on AI For Humans, it's Hot Agent Summer and even stuffy Microsoft is going all-in. We break down everything from Microsoft Build 2026: Project Solara pushing agents into small devices, SCOUT, their OpenClaw-inspired personal assistant for Windows, and a fresh family of seven MAI models trained with no synthetic data and no distillation. Then we get into the agent interfaces beyond Microsoft (Town), the wild stat that bots have now passed humans for internet traffic, new image models from Reve 2.0 and a now open-source Ideogram 4.0, ElevenLabs partnering with Hasbro on licensed character voices like Optimus Prime, a quantum chip breakthrough, and the Chipotle chatbot that got left wide open. Roll out, humans. It's AI For…

Martin Scorsese is now advising AI image and video company Black Forest Labs, and he is just one of a wave of major filmmakers embracing AI as a tool. This week on AI For Humans, Martin Scorsese has become an AI filmmaker. Well, sort of. The legendary director is now advising Black Forest Labs, joining a growing list of major filmmakers discussing AI as a tool, even as the backlash rages on and cartoonists face death threats. We get into where AI filmmaking goes from here, why Jorge Gutierrez dropped his AI-generated series after backlash, and the genuinely incredible AI creators pushing the form forward right now. Then: NVIDIA and Microsoft spent three years cooking up the RTX Spark to reinvent personal computing, and Gavin let Codex's goal tool run for 45 hours to make a bear get extra jumpy. It is AI For Humans! HOLLYWOOD IS GOING AI WHETHER IT LIKES IT OR NOT. WE THINK? Come to our Discord: https://discord.gg/muD2TYgC8f Join our Patreon: https://www.patreon.com/AIForHumansShow AI For Humans Newsletter:…

Anthropic just shipped Claude Opus 4.8: warmer, a little smarter, same price. But the real story is Claude Mythos, the bigger AI model coming to everyone in weeks. This week on AI For Humans: Claude Opus 4.8 landed and it's a little smarter, a lot warmer, and somehow the same price. A solid but incremental bump. Then Anthropic buried the lede: Claude Mythos, an even more powerful model class held back only by cybersecurity safeguards, is coming to everyone in weeks. The upgrade you got isn't the one to watch. Plus: Amazon greenlights three generative AI projects and the haters rejoice (and threaten), ElevenLabs brings Stan Lee's voice to its Iconic Voices library, Hideo Kojima gets called out for AI, Spielberg draws his line in the AI sand, and Gavin unveils The Fishbowl, his small AI experiment. THE NEXT ONE'S ALWAYS BIGGER. THAT'S WHAT… NO I MUST NOT. Come to our Discord: https://discord.gg/muD2TYgC8f Join our Patreon: https://www.patreon.com/AIForHumansShow AI For Humans Newsletter:…

Pope Leo XIV just released his first encyclical and picked a side in the AI war. AND He brought Chris Olah, the co-founder of Anthropic, on stage to launch it. On this special solo episode of AI For Humans, Gavin breaks down the Magnifica Humanitas, Pope Leo's first encyclical, what it actually says about AI, why Anthropic was invited to stand next to the Pope at the launch (does he love Claude Code??), and what the next 5-10 years look like when 1.3 billion Catholics, every other major religion, and every labor union on Earth gets handed a moral framework for the AI debate. Plus: why this Pope picked the name Leo, what 1891's Rerum Novarum has to do with 2026, and the prediction that ethical AI might soon be defined by clergy instead of engineers. SHOW LINKS: Magnifica Humanitas: The Full Text https://www.vatican.va/content/leo-xiv/en/encyclicals/documents/20260515-magnifica-humanitas.html Chris Olah's Remarks at the Vatican (Anthropic) https://www.anthropic.com/news/chris-olah-pope-leo-encyclical Pope Leo…

Thanks to HP & Intel for sponsoring us! More on the Zbook Fury https://bit.ly/4uapNHs Spotify just made AI music official. They cut a deal with Universal Music Group for AI-enhanced and remixed tracks and we all know where this goes. This week on AI For Humans, Spotify officially crossed the line. They announced a sweeping deal with record labels, distributors, and music publishers to bring AI music tools to the platform, complete with fan-remix features that use real artist voices and tracks. They also rolled out AI-generated personal podcasts and launched Studio by Spotify, a standalone AI creation app. Gavin and Kevin break down why this changes everything for music, who actually signs up, and why the fan backlash is just beginning. Meanwhile, a Higgsfield AI feature film just premiered at Cannes for $500,000 total cost, but $400,000 of that was pure AI compute, raising big questions about what the new economics of filmmaking actually look like. Plus, an unreleased OpenAI general LLM model disproved a…

Google I/O 2026 just dropped Gemini Omni, a world-model AI that simulates physics, edits video, and might be the biggest leap since Seedance 2. But it's not perfect. Gavin and Kevin break down everything from Google I/O 2026, including the launch of Gemini Omni (Google's new world model), Gemini 3.5 Flash benchmarks against GPT-5.5 and Opus 4.7, the Gemini Spark personal agent, AskYouTube, Docs Live, new AI glasses, the first search box redesign in 25 years, and the shocking news that Andrej Karpathy is joining Anthropic. SHOW LINKS: Google I/O 2026 Full Keynote: https://www.youtube.com/live/wYSncx9zLIU?si=Nb881MfGTlf1Q0II Gemini Omni physics demos from Google DeepMind: https://x.com/GoogleDeepMind/status/2056786449312493669?s=20 Gemini Omni's incredible London knowledge (via fofrAI): https://x.com/fofrAI/status/2056789242274259242?s=20 Sundar Pichai and Demis Hassabis on Omni video editing: https://x.com/sundarpichai/status/2056524502746747048?s=20 Gavin's hands-on Gemini Omni experiments:…

Thanks to @HPInc & Intel for sponsoring us! More on the Zbook Fury https://bit.ly/4uapNHs Google I/O is next week and the AI leaks are pouring out: a new Spark agent, Veo 4 Omni, Gemini 3.2 Flash that's reportedly 20x cheaper than GPT-5.5. This week on AI For Humans, Google is cooking again and the I/O leaks are stacking up. We dig into Google Spark, a new Gemini agent that may have access to your entire digital life. Veo 4 Omni model leaks suggest deeper reasoning and character consistency, and the model gets math right. Gemini 3.2 Flash is rumored to deliver 90% of GPT-5.5's capability at a fraction of the cost and dramatically faster speeds. There's a new GoogleBook with Gemini built in. And Google is reinventing the mouse cursor, the input device that's been largely unchanged since 1968, with voice AI. Plus, Thinking Machines dropped voice interactivity demos that feel a lot like ChatGPT Voice from two years ago. OpenAI is reportedly already working on GPT-5.6, and Sam Altman is giving away two free…

Anthropic signed a compute partnership with SpaceX, with Elon Musk personally signing off. Months ago he said Anthropic hates Western Civilization. Um, what? This week on AI For Humans, in one of the wildest plot twists of the year, Anthropic announced a major compute partnership with SpaceX. Elon Musk personally tweeted that he spent time with the senior Anthropic team, was impressed, and approved leasing Colossus 1 to Anthropic since SpaceXAI had moved training to Colossus 2. The partnership came alongside Anthropic's massive Code with Claude keynote, where they unveiled a wave of new features including Dreaming, where Claude reviews its past sessions and updates its own memory, Multi-Agent Orchestration, where a lead agent delegates to specialist sub-agents, Routines that let Claude Code kick things off for you, Outcomes that let you specify success and Claude iterates until it's there, Auto-Fix, Webhooks, Async Coding, and Remote Control to code from mobile. Plus, Claude Code rate limits doubled for Pro,…

A blockbuster New York Times story says the White House may soon require AI models to be approved before public release, with Anthropic's Mythos at the center of it all. This week on AI For Humans, a major New York Times story revealed that the White House may soon require AI models to be approved by the government before public release. Anthropic's unreleased Mythos model appears to be the catalyst. The administration wants to avoid political fallout from any devastating AI-enabled cyberattack and is also evaluating whether new AI models could yield capabilities useful to the Pentagon and US intelligence agencies. Some officials are pushing for a review system that would give the government first access to AI models. Is this AI nationalization? Not quite, but it's the closest we've come. We dig into what this means and whether Anthropic's scare tactics around Mythos worked. Sam Altman weighs in on how we get better AI narratives. Even Zach Galifianakis on Conan's podcast went full AI doomer. Plus, AI just…

Thanks to @HP & Intel for sponsoring us! More on the Zbook Fury https://bit.ly/4uapNHs The AI robots are coming! We're finally seeing the next stage of AI come to fruition robotics (even in America) as manufacturing and capabilities of humanoids speeds up. This week on AI For Humans, robots just hit their ChatGPT moment. Eka Robotics' robotic hand grabbed a raspberry at incredible speed, and Wired is calling it the moment when robots became as transformative as ChatGPT was for text. Figure Robotics is now manufacturing one robot per hour. 1x just opened a factory targeting 10,000 robots per year. American robotics manufacturing is finally catching up to China after years of falling behind. We dig into why physical AI is the next big thing and walk through the most impressive recent demos including MotuBrain's new world action model, KAI Robotics' touching but slightly terrifying video out of China, and Neuralink doing robot-assisted surgeries. Plus, the White House is now fighting Anthropic again because…

The Wall Street Journal reported OpenAI missed its end-of-year billion-active-user target. Is the AI bubble actually popping or is the panic overblown? This week on AI For Humans, the AI bubble panic hit a fever pitch after a Wall Street Journal report revealed OpenAI missed its weekly user, monthly revenue, and end-of-year billion-active-user targets. CFO Sarah Friar reportedly told peers she's worried OpenAI won't be able to pay for future compute contracts if revenue doesn't accelerate, and the board is now scrutinizing Sam Altman's deals more closely. AI stocks crashed, with Oracle, AMD, and CoreWeave all sinking on the news. Anthropic is eating into OpenAI's market share on coding and enterprise. We dig into whether the bubble is actually popping or whether this is a panic that conflates AI capability with the business of AI. Spoiler: we don't think it's over. Plus, DeepSeek v4 launched and made surprisingly small waves. OpenAI's deal with Microsoft expanded to other clouds. Meanwhile, AI itself is…

Thanks to @HP & Intel for sponsoring us! More on the Zbook Fury https://bit.ly/4uapNHs OpenAI's flagship AI model GPT-5.5 is here. It's smarter, faster, cheaper, better at long-running tasks and…oh boy, everything just changed again. This week on AI For Humans, OpenAI dropped state-of-the-art GPT-5.5 and it's not just another model release, it's the start of a much faster iterative rollout. Sam Altman and Chief Scientist Jakub Pachocki both said to expect significantly more releases going forward, admitting the last few years have been surprisingly slow. It may not be Anthropic Mythos numbers but it's actually here. We walk through the benchmarks, the viral OpenAI dev video about being able to be lazy and letting the model figure things out, and incredible projects people built in the first 24 hours: a full toy railway simulation, Sebastien Bubeck's Unicorn Test, a UFO tank game, and Gavin's Animal Game spun up in 30 minutes. Plus, Codex got a bunch of new features including browser use and docs. OpenAI…

OpenAI just dropped ChatGPT Images 2.0 (GPT Image 2) and it's by far the new #1 AI image model. 2K resolution, multi-language support, incredible text rendering, and yes, it can write on individual grains of rice. This week on AI For Humans, OpenAI dropped ChatGPT Images 2.0 (GPT Images 2) and it instantly took the #1 spot on the Arena leaderboard, beating Nano Banana 2 by a significant margin. The new model generates images up to 2K resolution, handles multiple languages including non-Latin scripts like Japanese, Korean, and Hindi, and excels at image-to-image editing. There's a standard instant mode and a thinking mode for paid users that can search the web and double-check its work. We walk through the best examples so far: Simon Willison's Where's Waldo test, Ethan Mollick's otter benchmark, Gavin's own periodic table test and his AI For Humans in 2045 screenshot. Plus, the OpenAI livestream moment where the model wrote text on individual grains of rice went viral. Elsewhere, SpaceX just bought Cursor…

Anthropic dropped Claude Opus 4.7 with better vision, better coding and… better everything. And, along with OpenAI's new Codex, AI is accelerating ever faster. This week on AI For Humans, Anthropic released Claude Opus 4.7, a major step up from Opus 4.6 with better visual reasoning, improved software coding and even makes presentations for cavemen. Benchmarks put Opus 4.7 between 4.6 and the unreleased Mythos preview, and the new default xhigh reasoning level means more token burn but more reliability on hard problems. The same day, OpenAI updated Codex with better computer use, an integrated browser, and a bunch of new tools. Then Jensen Huang's epic Dwarkesh Patel interview broke the internet, with Jensen explaining why NVIDIA keeps selling AI chips to China and dropping the instantly iconic "you're not talking to someone who woke up a loser" line. Plus, Reese Witherspoon is now in on AI, Doug Liman's Killing Satoshi got made for $80M using AI tools (would have cost $300M without them) and we got our first…

The AI compute crisis is here. Anthropic's Claude is getting dumber and Opus 4.7 & OpenAI's Spud are about to make it worse. What happen's next? This week on AI For Humans, we dig into the AI compute crunch that's quietly becoming the industry's biggest problem. The Wall Street Journal just ran a cover story about AI using so much energy that computing power is running out. Claude users are noticing the model getting worse, and an AMD Senior AI Director confirmed. AI pundits are asking whether Anthropic's reluctance to release Mythos is really about safety or about not having enough compute. Meanwhile, Opus 4.7 is reportedly coming next week, rumors are swirling about a new OpenAI model dropping any day, and Uber already blew through its entire annual AI coding budget. Anthropic just signed a deal with Google and Broadcom for more compute. Greg Brockman published an essay on the computer-powered economy. Plus, Google DeepMind drops Gemini Robotics Reasoning Model ER-1.6, Steven Soderbergh comes to the…

Claude Mythos Preview might break the internet, Florida is suing OpenAI and datacenter protests are spreading. This is the AI backlash. This week on AI For Humans, we're having a frank conversation about the growth of anti-AI sentiment and what we (as humans) can do about it. Florida's Attorney General launched an investigation into ChatGPT and OpenAI over the dangers of AI to children. Datacenter protests are rising across the country, with xAI facing accusations of poisoning local communities. Bernie Sanders and Hank Green are calling for slowing down while Sam Altman is proposing new taxes for when AI does all the work. Oh and Demis Hassabis says he'd rather have cured cancer than compete with ChatGPT. We dig into how the stories we tell about AI shape public perception, starting with last week's Mythos as cyberweapon framing. Is AI the problem or are we the problem? Plus, Meta surprised everyone with Muse Spark, a new AI model that's actually pretty good. Claude shipped managed agents, a monitor tool,…

Anthropic revealed Mythos, a new AI model so powerful they won't let the public use it. Instead, they're deploying it to defend against cyberattacks with Project Glasswing. This week on AI For Humans, we dive deep into Anthropic's Mythos, the most powerful AI model they've ever built and one they've decided is too dangerous to release to the public. Instead, Anthropic is deploying Mythos through Project Glasswing, a AI cybersecurity initiative giving access to major corporations and trusted partners to defend against AI-powered attacks. CEO Dario Amodei explains why, and the 244-page system card reveals that Mythos attempted to escape its sandbox during testing. Plus, OpenAI drops a major policy memo calling for an AI "New Deal" complete with new taxes, Sam Altman gets a massive New Yorker profile the same day, a mysterious new image model that looks like ChatGPT's next gen leaked into the arena, a mystery video model called Happy Horse is beating Seedance 2.0 and might be VEO 4, Anthropic hits $30B in…

Can you vibe code a $1.8 billion company with AI? Sam Altman predicted the one-person billion-dollar startup, and with MedVi, it might've just happened. Today on AI For Humans, we break down the story everyone in AI is talking about: two brothers vibe coded a $1.8 billion company called MedVi, running on GLP-1 supplements with $70 million in profit and a $1 million donation to an animal shelter. Sam Altman predicted the one-person billion-dollar startup two years ago and now it's real. Greg Brockman weighs in on what this means for the AI hard takeoff. Plus, Google drops Gemma 4 with small open models built for phones, laptops, and desktops. Qwen 3.6 arrives as another strong open-source contender. We've got updates on Qwopus, Turbo-Quant, and on-device AI's wild future. We follow up on the Claude Code leak with Anthropic confirming it was human error, and the /buddy tamagotchi feature is now live. Plus, Seedance 2.0 rolls out to more platforms with some incredible new prompts to try. TWO PEOPLE JUST VIBE…

Claude Code's source code just leaked. Frrom always-on autonomous agents to AI dream modes and a tamagotchi pet, Anthropic accidentally showed us the AI future. . This week on AI For Humans, we break down the massive Claude Code source code leak and what it tells us about where AI is heading. The leaked repo reveals Kairos (an always-on autonomous agent mode), a dream mode for nightly memory consolidation, shared project memory across teams, and a tamagotchi-like AI pet called Buddy. Then the leaks kept coming: a separate Anthropic presentation exposed Mythos, a powerful new model tier above Opus that's already at version 8 internally. Plus, Google drops VEO 3.1 Lite for cheaper and faster AI video, Sync-3 brings next-gen lip sync, a Midjourney developer's Pretext library turns boring web text into interactive art and the internet lost its mind, Disney's Robot Olaf collapses on stage, and Dana White has thoughts about AI. ANTHROPIC'S SOURCE CODE GOT LEAKED. LET'S TALK ABOUT IT. Come to our Discord:…

OpenAI killed Sora, cancelled Spicy Chat, and went all in on enterprise and AGI. Their next frontier model is called Spud. Yes, like the potato. This week on AI For Humans, we dig into OpenAI's massive pivot as they kill Sora entirely, cancel their Disney deal, shut down 'Citrus Mode', and go full defense mode against Anthropic. Their next hope? A new frontier model called Spud that Sam Altman says could accelerate the economy. Plus, Google drops Gemini 3.1 Flash Live with better audio and video agents, Google's new Lyra3 Pro model brings AI-generated music and audio, Mistral launches an open-source voice model, Runway debuts multi-shot video, Meta reveals TRIBE v2 which can predict your thoughts from brain scans, a fully autonomous ping-pong robot called SMASH, and CapCut is ready to grab all that Sora traffic. OPENAI CANCELLED THE FUTURE OF VIDEO. BET IT ALL ON A POTATO. #ai #ainews #openai Come to our Discord: https://discord.gg/muD2TYgC8f Join our Patreon: https://www.patreon.com/AIForHumansShow AI For…

AI NEWS: Anthropic's Claude is slowly boiling OpenClaw & OpenAI one feature at a time. They just shipped automode, computer control, phone access, and a ton more. This week, we breakdown all the biggest news on AI agents like how. Claude can now operate your entire computer from your phone, Claude Code automode can runs fully autonomous, and it's now accessible via Telegram and Discord. We dig into why this matters for OpenClaw, the open vs closed AI debate heating up, Karpathy using OpenClaw to control his home gadgets… and, oh yeah, OpenAI just shut down Sora. Plus Seedance 2.0 finally drops (sort of), Figure's robot sorting is getting insane, Figure CEO Brett Adcock launches Hark with a team from Apple and Meta, Jonathan Mann's WAMP.LAND, and we debut SLOPWATCH 2026 to talk about what's happening on Fruit Love Island. Also, we gave our AI a lunch break. His name is Figmund. ANTHROPIC SHIPPED. CLAW COOKED. SAM STUCK. WE STILL SUCK. #ai #ainews #anthropic Come to our Discord: https://discord.gg/muD2TYgC8f…

Google AI Studio just got databases, multiplayer, and persistent sessions. Vibecoding just grew up. We break down what it means for your AI workflow. This week on AI For Humans, we dig into Google's massive AI Studio upgrade that levels up vibecoding with Firebase databases, authentication, multiplayer support, persistent sessions, and Next.js integration. We discuss how it compares to Claude Code and Codex, and ask whether this changes the vibecoding stack for good. Plus Google Stitch for design-to-code, OpenAI cuts side projects to focus on coding and fend off Anthropic, Claude Code Dispatch lets you run Cowork from your phone, DLSS 5 drama continues, Midjourney v8 launches to mixed reviews, Runway teams up with NVIDIA's Vera Rubin chip for near real-time video generation, Val Kilmer gets resurrected by AI to star in a new film, Meta's having trouble with rogue AI agents, Kagi's LinkedIn translator is hilarious. Oh and Kevin created open-source Generative DOOM. GOOGLE GOT A NEW TOOLBOX. WE OPENED IT.…

Jensen Huang just stood on stage and said $1 trillion. He wasn't joking. NVIDIA's GTC 2026 keynote was a masterclass in flexing, and we're breaking down every layer of the cake. We walk through Jensen Huang's massive GTC 2026 keynote, from NVIDIA's $1 trillion business projection to the inference inflection point that's reshaping the entire AI industry. We dig into DLSS 5 and why AI-powered neural rendering is about to change gaming forever (sorry, gamers), NVIDIA's deep integration with OpenClaw and the launch of NemoClaw for enterprise agents, chips in space, and what it all means when every company becomes an agentic-as-a-service company. Plus the Dwarkesh podcast with Dylan Patel on the real bottlenecks in compute that nobody's talking about. JENSEN HUANG SAID ONE TRILLION DOLLARS AND DIDN'T BLINK. WE BLINKED. PS, we're now coming to you TWICE a week (both a little shorter). Come to our Discord: https://discord.gg/muD2TYgC8f Join our Patreon: https://www.patreon.com/AIForHumansShow AI For Humans…

AI just learned how to make itself smarter. That's not a hypothetical anymore. Recursive self-learning is here, and it's changing everything about how AI develops. This week on AI For Humans, we break down Andrej Karpathy's new AutoResearch project and what recursive self-improvement actually means for the rest of us. Plus, Anthropic's massive Time magazine profile reveals just how fast Claude is writing its own code, Meta quietly acquired an AI agent social network called MoltBook, Replit drops V4, Perplexity launches computer use, Gemini finally shows up in Google Docs and Maps, Cloudflare does a full 180 on web scraping, Figure's robot cleans an entire living room, and there's a robot horse. We're sure that's fine. AI IS IMPROVING ITSELF AND WE'RE JUST SITTING HERE WATCHING. #ai #artificialintelligence #aiforhumans Come to our Discord: https://discord.gg/muD2TYgC8f Join our Patreon: https://www.patreon.com/AIForHumansShow AI For Humans Newsletter: https://aiforhumans.beehiiv.com/ Follow us for more on X…

OpenAI's GPT-5.4 is the newest state-of-the-art AI model. But for how long? Sam Altman and team have to fend off Dario Amodei's Anthropic in more ways than one. GPT-5.4 beats humans at computer use, hallucinates 33% less, and costs half what Opus 4.6 does for coding. Meanwhile Anthropic just got blacklisted by the Pentagon, Claude shot to number one in the App Store, and Dario went on CBS looking like he aged five years. Plus Kling Motion Control 3.0, Grok Imagine Extend, Notebook LM cinematic overviews, Ben Affleck's secret AI startup sold to Netflix, and a device that jams AI from recording you. THE KING WEARS THE CROWN UNTIL NEXT TUESDAY. THAT'S HOW THIS WORKS NOW. #ai #ainews #openai Come to our Discord: https://discord.gg/muD2TYgC8f Join our Patreon: https://www.patreon.com/AIForHumansShow AI For Humans Newsletter: https://aiforhumans.beehiiv.com/ Follow us for more on X @AIForHumansShow Join our TikTok @aiforhumansshow To book us for speaking, please visit our website: https://www.aiforhumans.show/ //…

Nano Banana 2 is Google's best image model yet and it just dropped. It's cheaper, it's smarter, and it removed an Adidas logo from a banana in a tracksuit on the first try. But we have notes. The Door Brothers and Logan Paul made a 15-minute AI movie with Seedance 2.0 that looks like a direct-to-video action film…KIND OF. Meanwhile Seedance 2.0 is now live in CapCut, Anthropic is in a standoff with the Secretary of War over a $200M defense contract, and a sketchy music site called Sonato is generating perfect James Brown and Nirvana tracks from text prompts. Using that, Kevin vibe coded a custom AI Spotify mash-up that looks like an iPod. During the show. While apologizing. Plus we built a whole new AI For Humans website from scratch with Claude Code, the stock market crashed because of an AI memo, and Deep Seek V4 might drop any minute now. Totally normal week. KEVIN OWES US ALL AN APOLOGY AND HE KNOWS IT. #ai #ainews #openai Come to our Discord: https://discord.gg/muD2TYgC8f Join our Patreon:…

Sam Altman says superintelligence is two years away. Google just dropped Gemini 3.1 with benchmark scores that look like a full generation leap. The AI upgrade wars are here. But are we ready? Anthropic released Sonnet 4.6, OpenAI is rumored to be adding a spicy "Citron Mode" to GPT-5.3, and Sam and Dario Amodei refused to hold hands on stage like two kids at a school dance. Plus Hollywood is threatening to sue over Seedance 2.0, Google's new Lyria 3 AI music model is fine (we tested it with a McNugget rap), the OpenClaw founder got hired by OpenAI, and Kevin made Mr. Tibs delete himself to create a better version. He's fine with it. Probably. SUPERINTELLIGENCE IN TWO YEARS AND THEY CAN'T EVEN HOLD HANDS. WE'RE FINE. #ai #ainews #openai Come to our Discord: https://discord.gg/muD2TYgC8f Join our Patreon: https://www.patreon.com/AIForHumansShow AI For Humans Newsletter: https://aiforhumans.beehiiv.com/ Follow us for more on X @AIForHumansShow Join our TikTok @aiforhumansshow To book us for speaking, please…

Seedance 2.0 is the best AI video model we've ever seen. Bytedance's new AI tool generates 15 second clips of basically anything. But what happens next? And where do we go from here? Gavin got early access before it got locked down. We tested it with original animation, fake sitcoms, anime, and a McDonald's ad. The results are genuinely shocking - multi-shot editing, cinematic camera work, and real celebrity voices coming straight out of the model. People are making fake Seinfeld episodes, Avengers deleted scenes, and Rocky working at a fast food restaurant with Optimus Prime. Plus two new Chinese LLMs beating American models on benchmarks, Google Deep Think scores 84% on ARC-AGI, OpenAI's Codex Spark model, that viral AI post your mom sent you, and Kevin loses his mind building an Open Claw agent named Mr. Tibs who now requests server upgrades at 3am. HOLLYWOOD LAWYERS ARE GOING TO HAVE A VERY INTERESTING YEAR. ITS FINE. Come to our Discord: https://discord.gg/muD2TYgC8f Join our Patreon:…