Last Week in AI

Our 243rd episode with a summary and discussion of last week's big AI news! Recorded on 04/29/2026 Hosted by Andrey Kurenkov and Jeremie Harris Feel free to email us your questions and feedback at andreyvkurenkov@gmail.com and/or hello@gladstone.ai Read out our text newsletter and comment on the podcast at https://lastweekin.ai/ In this episode: Timestamps: OpenAI Unveils Its New, More Powerful GPT-5.5 Model - The New York TimesxAI Launches grok-voice-think-fast-1.0: Topping τ-voice Bench at 67.3%, Outperforming Gemini, GPT Realtime, and More - MarkTechPostClaude can now plug directly into Photoshop, Blender, and Ableton | The Verge China's DeepSeek releases preview of long-awaited V4 model as AI race intensifiesTencent Unveils Hy3 preview; Model Enhances Agent Capabilities and Real-World Usability - Tencent 腾讯ClawMark: A Living-World Benchmark for Multi-Turn, Multi-Day, Multimodal Coworker Agents Google Plans to Invest Up to $40 Billion in AnthropicMeta will use hundreds of thousands of AWS Graviton chipsChina blocks Meta's $2 billion takeover of AI startup ManusOpenAI shakes up partnership with Microsoft, capping revenue share paymentsElon Musk Testifies of AI Risk at Trial, Says OpenAI Tried to ‘Steal’ a Charity - WSJJudge rejects DOJ bid to delay Anthropic appeal in Pentagon disputeGoogle’s Gemini can now run on a single air-gapped server — and vanish when you pull the plugDeepMind's David Silver just raised $1.1B to build an AI that learns without human data | TechCrunch Evaluating whether AI models would sabotage AI safety researchLLMs Corrupt Your Documents When You DelegateTemporal Sparse Autoencoders: Leveraging the Sequential Nature of Language for InterpretabilityMemorandum on Adversarial Distillation of American AI ModelsTeen boys are dating their AI chatbots—and experts warn it could kill their careers | FortuneAnnouncing the Anthropic Economic Index SurveyScoop: CISA lacks access to Anthropic's Mythos Taylor Swift Files to Trademark Voice and Likeness to Protect Against AI Misuse Maximal Brain Damage Without Data or Optimization: Disrupting Neural Networks via Sign-Bit Flips See Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

Our 242nd episode with a summary and discussion of last week's big AI news! Recorded on 04/22/2026 Hosted by Andrey Kurenkov and Jeremie Harris Feel free to email us your questions and feedback at andreyvkurenkov@gmail.com and/or hello@gladstone.ai Read out our text newsletter and comment on the podcast at https://lastweekin.ai/ In this episode: Timestamps: ChatGPT's new Images 2.0 model is surprisingly good at generating text | TechCrunchAlibaba Drops Qwen 3.6 Max Preview—Its Most Powerful Model Yet - DecryptGoogle launches Deep Research and Deep Research Max agents to automate complex researchMozilla Used Anthropic’s Mythos to Find and Fix 271 Bugs in Firefox | WIREDOrdering with the Starbucks ChatGPT app was a true coffee nightmare | The Verge SpaceX is working with Cursor and has an option to buy the startup for $60B | TechCrunchAI chip startup Cerebras files for IPO | TechCrunchTwo startups want to replace how AI learns: one just raised $180M, another is seeking up to $1BMonths-old start-up Recursive Superintelligence raises $500mn for self-teaching AIAnthropic takes $5B from Amazon and pledges $100B in cloud spending in return | TechCrunchKevin Weil and Bill Peebles exit OpenAI as company continues to shed 'side quests' | TechCrunchMeta hires five Thinking Machines Lab founders including a reported $1.5 billion engineerMeta cuts 198 Bay Area jobs as even larger layoffs reportedly loomMeta employees are up in arms over a mandatory program to train AI on their mouse movements and keystrokesChinese fabs import record volumes of US chipmaking equipment via Singapore and Malaysia — homegrown tool makers booked record 2025 revenues as price competition squeezes marginsGoogle Eyes New Chips to Speed Up AI Results, Challenging NvidiaCanadian quantum company Xanadu soars to $16 billion valuation after Nvidia release Moonshot AI releases Kimi-K2.6 model with 1T parameters, attention optimizations - SiliconANGLEMiniMax Just Open Sourced MiniMax M2.7: A Self-Evolving Agent Model that Scores 56.22% on SWE-Pro and 57.0% on Terminal Bench 2 - MarkTechPost Infusion: Shaping Model Behavior by Editing Training Data via Influence FunctionsScoop: NSA using Anthropic's Mythos despite blacklistUnauthorized group has gained access to Anthropic’s exclusive cyber tool Mythos, report claims Parcae: Scaling Laws For Stable Looped Language ModelsOccuBench: Evaluating AI Agents on Real-World Professional Tasks via Language Environment Simulation Deezer says 44% of songs uploaded to its platform daily are AI-generated | TechCrunchCelebrities will be able to find and request removal of AI deepfakes on YouTube | The VergeSee Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

Our 241st episode with a summary and discussion of last week's big AI news! Recorded on 04/18/2026 Hosted by Andrey Kurenkov and Jeremie Harris Feel free to email us your questions and feedback at andreyvkurenkov@gmail.com and/or hello@gladstone.ai Read out our text newsletter and comment on the podcast at https://lastweekin.ai/ In this episode: Timestamps: Anthropic releases Claude Opus 4.7, narrowly retaking lead for most powerful generally available LLM | VentureBeatMeta debuts the Muse Spark model in a 'ground-up overhaul' of its AI | TechCrunchOpenAI Launches GPT-5.4-Cyber with Expanded Access for Security TeamsOpenAI’s big Codex update is a direct shot at Claude Code | The VergeAnthropic launches Claude Design, a new product for creating quick visualsAnthropic’s New Product Aims to Handle the Hard Part of Building AI Agents | WIREDCanva’s AI 2.0 update goes all in on prompt-powered design tools | The VergeAdobe’s new AI Assistant marks a ‘fundamental shift’ in creative work | The VergeGemini can now pull from Google Photos to generate personalized images | The VergeGoogle rolls out a native Gemini app for Mac | TechCrunchChrome now lets you turn AI prompts into repeatable ‘Skills’ | The Verge Anthropic loses appeals court bid to temporarily block Pentagon blacklistingJeff Bezos’ AI lab poaches xAI cofounder Kyle Kozic from OpenAI. | The VergePerplexity's Shift to AI Agents Boosts Revenue 50%Anthropic Agrees to Rent CoreWeave AI Capacity to Power ClaudeCanada’s Cohere, Germany’s Aleph Alpha reportedly in merger talksChatGPT has a new $100 per month Pro subscription | The VergeOpenAI has bought AI personal finance startup Hiro | TechCrunchAllbirds announced a switch from shoes to AI and its stock jumped 600 percent | The Verge HY-World 2.0: A Multi-Modal World Model for Reconstructing, Generating, and Simulating 3D WorldsLyra 2.0: Explorable Generative 3D Worlds Daniel Moreno-Gama is facing federal charges for attacking Sam Altman’s home and OpenAI’s HQ | The VergeDuo accused of shooting at Sam Altman’s house are freed; no charges filed The Iranian Lego AI video creators credit their virality to ‘heart’ | The VergeHundreds of Fake Pro-Trump Avatars Emerge on Social Media - The New York TimesThe AI images Trump can’t get enough of | Donald Trump | The GuardianAutomated Weak-to-Strong ResearcherReproducing steering against evaluation awareness in a large open-weight modelIran threatens ‘complete and utter annihilation’ of OpenAI's $30B Stargate AI data center in Abu Dhabi — regime posts video with satellite imagery of ChatGPT-maker's premier 1GW data centerWall Street Banks Try Out Anthropic’s Mythos as US UrgesSee Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

Our 240th episode with a summary and discussion of last week's big AI news! Recorded on 04/08/2026 (sorry I keep releasing stuff late, will get better with it soon!) Hosted by Andrey Kurenkov and Jeremie Harris Feel free to email us your questions and feedback at andreyvkurenkov@gmail.com and/or hello@gladstone.ai Read out our text newsletter and comment on the podcast at https://lastweekin.ai/ In this episode: Timestamps: Anthropic debuts ‘Project Glasswing’ and new AI model for cybersecurity | The VergeGemini Live gets ‘biggest upgrade yet’ with Gemini 3.1 Flash LiveAnthropic says Claude Code subscribers will need to pay extra for OpenClaw usage | TechCrunchOpenAI abandons yet another side quest: ChatGPT's erotic mode | TechCrunchMicrosoft takes on AI rivals with three new foundational models | TechCrunchSuno leans into customization with v5.5 | The VergeAnthropic announces deal with Google, Broadcom, says revenue has tripledSam Altman May Control Our Future—Can He Be Trusted? | The New YorkerOpenAI, Anthropic, Google Unite to Combat Model Copying in China - BloombergChinese chipmakers claim nearly half of local market as Nvidia's lead shrinksSoftBank secures $40 billion loan to boost OpenAI investmentsGranola raises $125M at $1.5B valuation for its AI note-taking app - SiliconANGLEAnthropic acquires stealth startup Coefficient Bio in $400M dealOpenAI acquires TBPN, the buzzy founder-led business talk show | TechCrunchZ.AI Introduces GLM-5.1: An Open-Weight 754B Agentic Model That Achieves SOTA on SWE-Bench Pro and Sustains 8-Hour Autonomous Execution - MarkTechPostGoogle announces Gemma 4 open AI models, switches to Apache 2.0 license - Ars TechnicaZ.ai Launches GLM-5V-Turbo: A Native Multimodal Vision Coding Model Optimized for OpenClaw and High-Capacity Agentic Engineering Workflows EverywhereJudge blocks Pentagon’s effort to ‘punish’ Anthropic by labeling it a supply chain riskEmotion concepts and their function in a large language modelChina bars Manus co-founders from leaving country amid Meta deal review, FT reportsUS lawmakers ask whether Nvidia CEO's smuggling remarks misled regulatorsHow far does alignment midtraining generalize?Metagaming matters for training, evaluation, and oversightIran says it has struck Oracle data center in Dubai, Amazon data center in Bahrain — country has threatened to attack Nvidia, Intel, and others, tooSee Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

Our 239th episode with a summary and discussion of last week's big AI news! FYI: this one has pretty out of date news, I was traveling last week and failed to upload... apologies. Recorded on 03/25/2026 Hosted by Andrey Kurenkov and Jeremie Harris Feel free to email us your questions and feedback at andreyvkurenkov@gmail.com and/or hello@gladstone.ai Read out our text newsletter and comment on the podcast at https://lastweekin.ai/ In this episode: Timestamps: OpenAI Discontinues Sora App, Shuts Down Video Generation Service and API - BloombergAnthropic’s Claude Code and Cowork can control your computer | The VergeGemini task automation is slow, clunky, and super impressive | The VergeCursor Launches Composer 2 AI Model to Challenge OpenAI & AnthropicAdobe’s AI image generator can now be trained on your own art | The VergeLuma AI launches Uni-1, a model that outscores Google and OpenAI while costing up to 30 percent less | VentureBeatTrump Contracting Clause Would Override AI SafeguardsMeta accelerates AI ASIC roll-out as Broadcom secures four-generation chip design dealMicron revenue almost triples, tops estimates as demand for memory soarsElon Musk Unwraps $25 Billion Terafab Chip-Building Project - CNETZoox to widen US robotaxi footprint with San Francisco, Vegas expansionWaymo hits 170 million miles while avoiding serious mayhem | The VergeThe White House just laid out how it wants to regulate AI | CNN BusinessHow we monitor internal coding agents for misalignmentIncomplete Tasks Induce Shutdown Resistance in Some Frontier LLMsSummary: Mechanisms to Verify International Agreements about AI DevelopmentScoop: Anthropic meets with House Homeland Security behind closed doorsConsciousness Cluster: Preferences of Models that Claim they are ConsciousHyperAgentsSee Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

Our 238th episode with a summary and discussion of last week's big AI news! Recorded on 03/18/2026 Hosted by Andrey Kurenkov and Jeremie Harris Feel free to email us your questions and feedback at andreyvkurenkov@gmail.com and/or hello@gladstone.ai Read out our text newsletter and comment on the podcast at https://lastweekin.ai/ In this episode: * OpenAI released GPT-5.4 mini and nano with 400k-token context windows, higher per-token prices but claimed token-efficiency gains in Codex; nano is API-only and pitched for high-volume classification/data extraction despite a major price increase. * Mistral open-sourced the Small 4 model family (MoE, 119B total/6B active) combining reasoning, multimodal, and coding-agent capabilities, and announced Forge to help businesses train or post-train custom models. * Agent “operating system” competition intensified with Meta’s acquired Manus launching a local Mac agent, Nvidia announcing NeMo/“Open Shell” sandboxed agent runtime, and Nvidia also unveiling DLSS 5 plus major hardware forecasts including Groq LPU integration. * Business and safety updates included OpenAI shifting focus toward productivity/enterprise amid competition, Microsoft reorganizing Copilot and frontier-model efforts, Meta delaying its next model, China-linked ByteDance deploying large Nvidia clusters abroad, and new safety work on steganography, chain-of-thought faithfulness, fine-tuning defenses, cyber-attack evals, and constitution/spec compliance. A thank you to our current sponsors: odsc.ai/eastfactormeals.com/lwai50off Timestamps: OpenAI ships GPT-5.4 mini and nano, faster and more capable but up to 4x pricierMistral's new Small 4 model punches above its weight with 128 expert modulesMeta's Manus launches 'My Computer' to turn your Mac into an AI agent - 9to5MacNVIDIA Announces NemoClaw for the OpenClaw Community | NVIDIA NewsroomNvidia boosts knowledge work with Open Agent Development PlatformDLSS 5 looks like a real-time generative AI filter for video games | The VergeOpenAI to Launch ChatGPT 'Adult Mode' Despite Warnings From Its Own Advisers - CNETOpenAI Reportedly Pivoting to a Focus on Business and Productivity OnlyNvidia GTC 2026: CEO Jensen Huang sees $1 trillion in orders for Blackwell and Vera Rubin through ’27Mistral launches Forge to help enterprises build their own AI modelsChina's ByteDance gets access to top Nvidia AI chips, WSJ reportsMeta Delays Rollout of New A.I. Model After Performance ConcernsMicrosoft Shakes Up AI Division As Copilot Falls Behind Google and OpenAIA Decision-Theoretic Formalisation of Steganography With Applications to LLM MonitoringReasoning Theater: Disentangling Model Beliefs from Chain-of-ThoughtIn-Training Defenses against Emergent Misalignment in Language ModelsHow do frontier AI agents perform in multi-step cyber-attack scenarios?Eval awareness in Claude Opus 4.6’s BrowseComp performanceIntroducing Bloom: an open source tool for automated behavioral evaluationsHow well do models follow their constitutions?Nvidia’s H200 License Stirs Security Concern Among Top Democrats[2603.15031] Attention ResidualsMamba-3: Improved Sequence Modeling using State Space Principles See Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

Our 237th episode with a summary and discussion of last week's big AI news! Recorded on 03/13/2026 Hosted by Andrey Kurenkov and Jeremie Harris Feel free to email us your questions and feedback at andreyvkurenkov@gmail.com and/or hello@gladstone.ai Read out our text newsletter and comment on the podcast at https://lastweekin.ai/ In this episode: * Perplexity announced “Personal Computer,” a local Mac-based AI agent positioned as a safer alternative to OpenAI’s computer-use agents, while Anthropic added GitHub PR code review pricing reviews at $15–$25 and Cursor launched trigger-based “Automations” for always-on coding agents. * ChatGPT introduced interactive math/science visuals and Anthropic added in-chat interactive charts/diagrams; Nvidia released open weights for its 120B-parameter Natron Free Super hybrid Transformer–Mamba latent-MoE model trained natively at 4-bit for Blackwell GPUs. * Nvidia halted H200 production for China amid customs blocks and domestic chip pressure; xAI saw major co-founder departures; Anthropic previewed a Claude Marketplace for enterprise procurement; Yann LeCun’s aMI raised $1.3B; humanoid robot maker Sanctuary reached a $1.15B valuation. * Anthropic sued the Pentagon over a “supply chain risk” designation as memos ordered removal within 180 days; research covered models resisting activation steering, limits of chain-of-thought control, inference-scaling boosting cyber-task success, low-probability risky actions, weaknesses in SWE-bench, multimodal pretraining, long-context RNN memory caching, context-parallel training efficiency, RL for CUDA kernel optimization, and latent introspection detecting concept injection. A thank you to our current sponsors: odsc.ai/eastfactormeals.com/lwai50off Timestamps: Perplexity’s Personal Computer turns your spare Mac into an AI agent | The VergeAnthropic launches code review tool to check flood of AI-generated code | TechCrunchCursor is rolling out a new kind of agentic coding tool | TechCrunchChatGPT can now create interactive visuals to help you understand math and science concepts | TechCrunchAnthropic’s Claude AI can respond with charts, diagrams, and other visuals now | The Verge Introducing Nemotron 3 Super: An Open Hybrid Mamba-Transformer MoE for Agentic Reasoning | NVIDIA Technical Blog Nvidia halts H200 production as China backs Huawei AI chipsAnother XAI Cofounder Has Left, and Another Says He's Leaving. - Business InsiderAnthropic's Claude Marketplace allows customers to buy third-party cloud services | TechRadarYann LeCun's AMI Labs raises $1.03 billion to build world models | TechCrunchHumanoid robotics maker Sunday reaches $1.15B valuation to build household robots | TechCrunch Anthropic Sues Department of Defense Over ‘Supply Chain Risk’ Label - The New York TimesGoogle and OpenAI Just Filed a Legal Brief in Support of AnthropicInternal Pentagon memo orders military commanders to remove Anthropic AI technology from key systems - CBS NewsEndogenous Resistance to Activation Steering in Language ModelsReasoning Models Struggle to Control their Chains of Thought‘It means missile defence on datacentres’: drone strikes raise doubts over Gulf as AI superpowerEvidence for inference scaling in AI cyber tasks: Increased evaluation budgets reveal higher success ratesFrontier Models Can Take Actions at Low Probabilities Research note: Many SWE-bench-Passing PRs Would Not Be Merged into Main[2603.03276] Beyond Language Modeling: An Exploration of Multimodal PretrainingMemory Caching: RNNs with Growing MemoryUntied Ulysses: Memory-Efficient Context Parallelism via Headwise ChunkingCUDA Agent: Large-Scale Agentic RL for High-Performance CUDA Kernel GenerationLatent Introspection: Models Can Detect Prior Concept InjectionsPhysics of RL: Toy scaling laws for the emergence of reward-seekingSee Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

Our 236th episode with a summary and discussion of last week's big AI news! Recorded on 03/06/2026 Hosted by Andrey Kurenkov and Jeremie Harris Feel free to email us your questions and feedback at andreyvkurenkov@gmail.com and/or hello@gladstone.ai Read out our text newsletter and comment on the podcast at https://lastweekin.ai/ In this episode: * OpenAI released GPT-5.4 Pro with a 1M-token context window, mid-response course correction, native computer-use capabilities, improved tool use, higher GPT-VAL performance (83%), and “high cyber capability” safety measures; OpenAI also launched GPT-5.3 Instant with a less “preachy” tone and a claimed 26.8% hallucination reduction. * Google upgraded Gemini 3.1 Flash Lite with faster time-to-first-token and higher throughput, released a CLI for integrating agents with Gmail/Drive/Docs, and discussion highlighted real-world agent failure risks (including an example of an AI-driven mass email deletion). * Luma launched unified multimodal models and Luma Agents for end-to-end creative work across text, image, video, and audio, including a reported ad localization use case completed in 40 hours for under $20,000. * Defense-contract controversy escalated: Anthropic was labeled a supply chain risk (later narrowed), OpenAI’s DoD contract language emphasized “all lawful uses,” consumer cancellations boosted Claude’s app rankings, OpenAI saw departures and announced a $110B raise at a $730B valuation, Alibaba lost key Qwen leaders, a lawsuit alleged Gemini contributed to a suicide, Anthropic warned of major labor disruption, and METR corrected its AI time-horizon estimates. A thank you to our current sponsors: odsc.ai/eastfactormeals.com/lwai50off Timestamps: OpenAI launches GPT-5.4 with Pro and Thinking versions | TechCrunchOpenAI GPT-5.3 Instant less likely to beat around the bush • The RegisterGoogle releases Gemini 3.1 Flash Lite at 1/8th the cost of Pro | VentureBeatGoogle makes Gmail, Drive, and Docs 'agent-ready' for OpenClaw | PCWorldLuma launches creative AI agents powered by its new ‘Unified Intelligence’ models | TechCrunch Anthropic CEO Dario Amodei calls OpenAI's messaging around military deal 'straight up lies,' report says | TechCrunchNo ethics at all': the 'cancel ChatGPT' trend is growing after OpenAI signs a deal with the US military | TechRadarOpenAI raises $110B in one of the largest private funding rounds in history | TechCrunchAlibaba scrambles after sudden departure of Qwen tech lead Pentagon approves OpenAI safety red lines after dumping AnthropicWhere things stand with the Department of War AnthropicMicrosoft says Anthropic’s products remain available to customers after Pentagon blacklistA new lawsuit claims Gemini assisted in suicide | SemaforAnthropic just mapped out which jobs AI could potentially replace. A 'Great Recession for white-collar workers' is absolutely possible | FortuneWe're correcting a mistake in our modeling that inflated recent 50%-time horizons by 10-20%See Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

Our 235th episode with a summary and discussion of last week's big AI news! Recorded on 02/27/2026 Hosted by Andrey Kurenkov and Jeremie Harris Feel free to email us your questions and feedback at andreyvkurenkov@gmail.com and/or hello@gladstone.ai Read out our text newsletter and comment on the podcast at https://lastweekin.ai/ In this episode: A thank you to our current sponsors: odsc.ai/eastfactormeals.com/lwai50off Timestamps: Anthropic releases Sonnet 4.6 | TechCrunchGoogle Rolls Out Latest AI Model, Gemini 3.1 Pro - CNETElon Musk says Grok 4.20 public beta is now available: Capabilities of AI chatbot offered by xAI - The Times of IndiaAnthropic just released a mobile version of Claude Code called Remote Control | VentureBeatPerplexity announces "Computer," an AI agent that assigns work to other AI agents - Ars TechnicaMeta strikes up to $100B AMD chip deal as it chases 'personal superintelligence' | TechCrunchNvidia challenger AI chip startup MatX raised $500M | TechCrunchWorld Labs lands $1B, with $200M from Autodesk, to bring world models into 3D workflows | TechCrunchSimile Raises $100 Million for AI Aiming to Predict Human BehaviorStargate AI data centers for OpenAI reportedly delayed by squabbles between partners — sources say OpenAI, Oracle, and SoftBank disagreed on who would have ultimate control of the planned data centersChina to increase leading-edge chip output by 5x in two years, report claims — aims to lift 7nm and 5nm production to 100,000 wafers per month, targeting half a million monthly by 2030On Surprising Effectiveness of Masking Updates in Adaptive OptimizersThink Deep, Not Just Long: Measuring LLM Reasoning Effort via Deep-Thinking Tokensmodels have some pretty funny attractor statesWhen Models Manipulate Manifolds: The Geometry of a Counting TaskBRIDGE: Predicting Human Task Completion Time From Model PerformanceNESSiE: The Necessary Safety Benchmark -- Identifying Errors that should not ExistThe least understood driver of AI progressThe Persona Selection Model: Why AI Assistants might Behave like HumansAnthropic CEO Amodei says Pentagon's threats 'do not change our position' on AIMusk's xAI, Pentagon reach deal to use Grok in classified systemsDetecting and preventing distillation attacksOpenAI details expanding efforts to disrupt malicious use of AI in new report - SiliconANGLESee Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

Our 235th episode with a summary and discussion of last week's big AI news! Recorded on 01/02/2026 Hosted by Andrey Kurenkov and Jeremie Harris Feel free to email us your questions and feedback at contact@lastweekinai.com and/or hello@gladstone.ai Read out our text newsletter and comment on the podcast at https://lastweekin.ai/ In this episode: * Major model launches include Anthropic’s Opus 4.6 with a 1M-token context window and “agent teams,” OpenAI’s GPT-5.3 Codex and faster Codex Spark via Cerebras, and Google’s Gemini 3 Deep Think posting big jumps on ARC-AGI-2 and other STEM benchmarks amid criticism about missing safety documentation. * Generative media advances feature ByteDance’s Seedance 2.0 text-to-video with high realism and broad prompting inputs, new image models Seedream 5.0 and Alibaba’s Qwen Image 2.0, plus xAI’s Grok Imagine API for text/image-to-video. * Open and competitive releases expand with Zhipu’s GLM-5, DeepSeek’s 1M-token context model, Cursor Composer 1.5, and open-weight Qwen3 Coder Next using hybrid attention aimed at efficient local/agentic coding. * Business updates include ElevenLabs raising $500M at an $11B valuation, Runway raising $315M at a $5.3B valuation, humanoid robotics firm Apptronik raising $935M at a $5.3B valuation, Waymo announcing readiness for high-volume production of its 6th-gen hardware, plus industry drama around Anthropic’s Super Bowl ad and departures from xAI. Timestamps: See Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

Our 233rd episode with a summary and discussion of last week's big AI news! Recorded on 01/30/2026 Hosted by Andrey Kurenkov and Jeremie Harris Feel free to email us your questions and feedback at contact@lastweekinai.com and/or hello@gladstone.ai Read out our text newsletter and comment on the podcast at https://lastweekin.ai/ In this episode: Timestamps: Google adds Gemini AI-powered ‘auto browse’ to Chrome | The VergeUsers flock to open source Moltbot for always-on AI, despite major risks - Ars TechnicaGoogle Brings Genie 3 'World Building' Experiment to AI Ultra Subscribers - CNETOpenAI’s ChatGPT translator challenges Google Translate | The VergeOpenAI launches Prism, a new AI workspace for scientists | TechCrunch Exclusive: China gives nod to ByteDance, Alibaba and Tencent to buy Nvidia's H200 chips - sources | ReutersAI chip startup Ricursive hits $4B valuation 2 months after launchAI Startup Recursive in Funding Talks at $4 Billion Valuation - BloombergFlapping Airplanes and the promise of research-driven AI | TechCrunchFrom invisibility cloaks to AI chips: Neurophos raises $110M to build tiny optical processors for inferencing | TechCrunch Qwen3-Max-Thinking debuts with focus on hard math, codeChina's Moonshot releases a new open-source model Kimi K2.5 and a coding agent | TechCrunchAi2 launches family of open-source AI developer agents that adapt to any codebase - SiliconANGLETiny startup Arcee AI built a 400B-parameter open source LLM from scratch to best Meta’s Llama Post-LayerNorm Is Back: Stable, ExpressivE, and Deep[2601.19897] Self-Distillation Enables Continual Learning[2601.20802] Reinforcement Learning via Self-DistillationTeaching Models to Teach Themselves: Reasoning at the Edge of Learnability Amodei, Hoffman Join Tech Workers Decrying Minnesota Violence - BloombergSee Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

Our 232st episode with a summary and discussion of last week's big AI news! Recorded on 01/23/2026 Hosted by Andrey Kurenkov and Jeremie Harris Feel free to email us your questions and feedback at contact@lastweekinai.com and/or hello@gladstone.ai Read out our text newsletter and comment on the podcast at https://lastweekin.ai/ In this episode: Timestamps: OpenAI to test ads in ChatGPT as it burns through billions - Ars TechnicaOpenAI is launching age prediction for ChatGPT accountsGoogle now offers free SAT practice exams, powered by Gemini | TechCrunchBaidu’s AI Assistant Reaches Milestone of 200 Million Monthly Active Users - WSJThe Drama at Thinking Machines, a New A.I. Start-Up, Is Riveting Silicon Valley - The New York TimesZhipu AI breaks US chip reliance with first major model trained on Huawei stack | South China Morning PostElon Musk’s xAI launches world’s first Gigawatt AI supercluster to rival OpenAI and AnthropicSequoia to invest in Anthropic, breaking VC taboo on backing rivals: FTHumans&, a 'human-centric' AI startup founded by Anthropic, xAI, Google alums, raised $480M seed round | TechCrunchBlack Forest Labs Releases FLUX.2 [klein]: Compact Flow Models for Interactive Visual Intelligence - MarkTechPost[2601.10611] Molmo2: Open Weights and Data for Vision-Language Models with Video Understanding and Grounding[2601.10547] HeartMuLa: A Family of Open Sourced Music Foundation Models[2601.11044] AgencyBench: Benchmarking the Frontiers of Autonomous Agents in 1M-Token Real-World ContextsSTEM: Scaling Transformers with Embedding ModulesReasoning Models Generate Societies of ThoughtWhy LLMs Aren't Scientists Yet: Lessons from Four Autonomous Research AttemptsSenate passes bill letting victims sue over Grok AI explicit imagesBuilding Production-Ready Probes For GeminiAnthropic Publishes Claude AI's New Constitution | TIMEArtists Launch Stealing Isn't Innovation Campaign To Protest Big Tech See Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

Our 231st episode with a summary and discussion of last week's big AI news! Recorded on 01/16/2026 Hosted by Andrey Kurenkov and Jeremie Harris Feel free to email us your questions and feedback at contact@lastweekinai.com and/or hello@gladstone.ai Read out our text newsletter and comment on the podcast at https://lastweekin.ai/ In this episode: Timestamps: Anthropic’s new Cowork tool offers Claude Code without the code | TechCrunchGoogle’s Gemini AI will use what it knows about you from Gmail, Search, and YouTube | The VergeGoogle removes some AI health summaries after investigation finds “dangerous” flaws - Ars TechnicaGmail is getting a Gemini AI overhaulSlackbot is an AI agent now | TechCrunchAnthropic Raising $10 Billion at $350 Billion ValueElon Musk xAI raises $20 billion from Nvidia, Cisco, investorsNVIDIA Needs a Supply Chain ‘Miracle’ From TSMC as China’s H200 AI Chip Orders Overwhelm Supply, Triggering a BottleneckOpenAI signs deal, worth $10B, for compute from Cerebras | TechCrunchCoreWeave in focus as it amends credit agreementLMArena lands $1.7B valuation four months after launching its product | TechCrunchNemotron-Cascade: Scaling Cascaded Reinforcement Learning for General-Purpose Reasoning ModelsmHC: Manifold-Constrained Hyper-ConnectionsIQuest_Coder_Technical_ReportTII Abu-Dhabi Released Falcon H1R-7B: A New Reasoning Model Outperforming Others in Math and Coding with only 7B Params with 256k Context Window - MarkTechPostDeep Delta LearningRecursive Language ModelsConditional memory via scalable lookupExtending the Context of Pretrained LLMs by Dropping their Positional EmbeddingsConstitutional Classifiers++: Efficient Production-Grade Defenses against Universal JailbreaksNvidia CEO says purchase orders, not formal declaration, will signal Chinese approval of H200China AI Leaders Warn of Widening Gap With US After $1B IPO WeekJake Sullivan is furious that Trump removed Biden’s AI chip export controls | The VergeSee Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

Our 230th episode with a summary and discussion of last week's big AI news! Recorded on 01/02/2026 Hosted by Andrey Kurenkov and Jeremie Harris Feel free to email us your questions and feedback at contact@lastweekinai.com and/or hello@gladstone.ai Read out our text newsletter and comment on the podcast at https://lastweekin.ai/ In this episode: Timestamps: OpenAI bets big on audio as Silicon Valley declares war on screens | TechCrunch Nvidia buying AI chip startup Groq for about $20 billion, biggest dealExclusive | Meta Buys AI Startup Manus, Adding Millions of Paying Users - WSJCursor continues acquisition spree with Graphite deal | TechCrunchMicron Hikes CapEx to $20B with 2026 HBM Supply Fully Booked; HBM4 Ramps 2Q26Chinese fabs are reportedly upgrading older ASML DUV lithography chipmaking machines — secondary channels and independent engineers used to soup up Twinscan NXT series Z.AI launches GLM-4.7, new SOTA open-source model for codingEvaluating AI’s ability to perform scientific research tasks Large Causal Models from Large Language ModelsUniversally Converging Representations of Matter Across Scientific Foundation ModelsMETA-RL INDUCES EXPLORATION IN LANGUAGE AGENTSAre the Costs of AI Agents Also Rising Exponentially?METR eval for Opus 4.5How to game the METR plot New York governor Kathy Hochul signs RAISE Act to regulate AI safety | TechCrunchActivation Oracles: Training and Evaluating LLMs as General-Purpose Activation ExplainersMonitoring MonitorabilitySam Altman is hiring someone to worry about the dangers of AI | The VergeX users asking Grok to put this girl in bikini, Grok is happy obliging - India TodaySee Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

Our 229th episode with a summary and discussion of last week's big AI news! Recorded on 12/19/2025 Hosted by Andrey Kurenkov and Jeremie Harris Feel free to email us your questions and feedback at contact@lastweekinai.com and/or hello@gladstone.ai Read out our text newsletter and comment on the podcast at https://lastweekin.ai/ In this episode: Timestamps: Google launches Gemini 3 Flash, makes it the default model in the Gemini app | TechCrunchChatGPT launches an app store, lets developers know it's open for business | TechCrunchIntroducing GPT-5.2-Codex | OpenAIMeta partners with ElevenLabs to power AI audio across Instagram, Horizon - The Economic Times OpenAI to End Equity Vesting Period for Employees, WSJ SaysHow China built its ‘Manhattan Project’ to rival the West in AI chipsChina’s Huawei, SMIC Make Progress With Chips, Report FindsOpenAI in Talks to Raise At Least $10 Billion From Amazon and Use Its AI ChipsAmazon has a new leader for its ‘AGI’ group as it plays catch-up on AI | The VergeBroadcom reveals its mystery $10 billion customer is AnthropicVibe-coding startup Lovable raises $330M at a $6.6B valuation | TechCrunchFal nabs $140M in fresh funding led by Sequoia, tripling valuation to $4.5B | TechCrunch Nvidia Becomes a Major Model Maker With Nemotron 3 | WIREDMeta introduces new SAM AI able to isolate and edit audio • The Register[2512.14856] T5Gemma 2: Seeing, Reading, and Understanding LongerAnthropic makes agent Skills an open standard - SiliconANGLE Budget-Aware Tool-Use Enables Effective Agent ScalingRethinking Thinking Tokens: LLMs as Improvement OperatorsWhat if AI capabilities suddenly accelerated in 2027? How would the world know? Update to GPdfT-5 System Card: GPT-5.2Neural Chameleons: Language Models Can Learn to Hide Their Thoughts from Unseen Activation MonitorsAsync Control: Stress-testing Asynchronous Control Measures for LLM AgentsGoogle is powering a new US military AI platform | The VergeSee Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

Our 228th episode with a summary and discussion of last week's big AI news! Recorded on 12/12/2025 Hosted by Andrey Kurenkov and Jeremie Harris Feel free to email us your questions and feedback at contact@lastweekinai.com and/or hello@gladstone.ai Read out our text newsletter and comment on the podcast at https://lastweekin.ai/ In this episode: Timestamps: GPT-5.2 is OpenAI’s latest move in the agentic AI battle | The VergeRunway releases its first world model, adds native audio to latest video model | TechCrunchGoogle says it will link to more sources in AI Mode | The VergeChatGPT can now use Adobe apps to edit your photos and PDFs for free | The VergeTencent releases Hunyuan 2.0 with 406B parametersChina set to limit access to Nvidia’s H200 chips despite Trump export approvalDisney investing $1 billion in OpenAI, will allow characters on SoraUnconventional AI confirms its massive $475M seed roundSlack CEO Denise Dresser to join OpenAI as chief revenue officer | TechCrunchThe state of enterprise AI[2512.10791] The FACTS Leaderboard: A Comprehensive Benchmark for Large Language Model FactualityClaude 4.5 Opus' Soul Document[2512.08296] Towards a Science of Scaling Agent SystemsEvaluating Gemini Robotics Policies in a Veo World SimulatorGuided Self-Evolving LLMs with Minimal Human SupervisionMartingale Score: An Unsupervised Metric for Bayesian Rationality in LLM Reasoning[2512.07783] On the Interplay of Pre-Training, Mid-Training, and RL on Reasoning Language ModelsStabilizing Reinforcement Learning with LLMs: Formulation and PracticesGoogle’s AI unit DeepMind announces UK 'automated research lab'Trump Moves to Stop States From Regulating AI With a New Executive Order - The New York Times[2512.09742] Weird Generalization and Inductive Backdoors: New Ways to Corrupt LLMsForecasting AI Time Horizon Under Compute SlowdownsAI Security Institute focuses on AI measurements and evaluationsNvidia AI Chips to Undergo Unusual U.S. Security Review Before Export to ChinaU.S. Authorities Shut Down Major China-Linked AI Tech Smuggling NetworkRSL 1.0 has arrived, allowing publishers to ask AI companies pay to scrape content | The Verge See Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

Our 227th episode with a summary and discussion of last week's big AI news! Recorded on 12/05/2025 Hosted by Andrey Kurenkov and Jeremie Harris Feel free to email us your questions and feedback at contact@lastweekinai.com and/or hello@gladstone.ai Read out our text newsletter and comment on the podcast at https://lastweekin.ai/ In this episode: Timestamps: Deepseek 3.2 : New AI Model is Faster, Cheaper and SmarterBlack Forest Labs launches Flux.2 AI image models to challenge Nano Banana Pro and MidjourneySora and Nano Banana Pro throttled amid soaring demand | The VergeMistral closes in on Big AI rivals with new open-weight frontier and small models | TechCrunchKling's Video O1 launches as the first all-in-one video model for generation and editingRunway rolls out Gen 4.5 AI video model that beats Google, OpenAINVIDIA’s Partners Are Beginning to Tilt Toward Google’s TPU Ecosystem, with Foxconn Reportedly Securing TPU Rack OrdersAmazon releases an impressive new AI chip and teases an Nvidia-friendly roadmap | TechCrunchOpenAI declares ‘code red’ as Google catches up in AI race | The VergeAnthropic reportedly preparing for massive IPO in race with OpenAI: FTBlack Forest Labs raises $300M at $3.25B valuation | TechCrunchParis-based AI voice startup Gradium nabs $70M seed | TechCrunchOpenAI announced a 1 GW Stargate cluster in Abu DhabiOpenAI’s investment into Thrive Holdings is its latest circular dealOpenAI to acquire Neptune, an AI model training assistance startupAnthropic acquires developer tool startup Bun to scale AI codingMicrosoft drops AI sales targets in half after salespeople miss their quotas - Ars Technica[2511.22570] DeepSeekMath-V2: Towards Self-Verifiable Mathematical ReasoningEvo-Memory: Benchmarking LLM Agent Test-time Learning with Self-Evolving MemoryNested Learning: The Illusion of Deep Learning ArchitectureMulti-Agent Deep Research: Training Multi-Agent Systems with M-GRPOState of AI: An Empirical 100 Trillion Token Study with OpenRouterTrump signs executive order launching Genesis Mission AI projectOpenAI has trained its LLM to confess to bad behavior | MIT Technology ReviewUS senators seek to block Nvidia sales of advanced chips to ChinaSee Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

Our 226th episode with a summary and discussion of last week's big AI news! Recorded on 11/24/2025 Hosted by Andrey Kurenkov and co-hosted by Michelle Lee Feel free to email us your questions and feedback at contact@lastweekinai.com and/or hello@gladstone.ai Read out our text newsletter and comment on the podcast at https://lastweekin.ai/ In this episode: Timestamps: Google launches Gemini 3 with new coding app and record benchmark scores | TechCrunchGoogle launches Nano Banana Pro powered by Gemini 3Anthropic releases Opus 4.5 with new Chrome and Excel integrations | TechCrunchOpenAI releases GPT-5.1-Codex-Max to handle engineering tasks that span twenty-four hoursChatGPT launches group chats globally | TechCrunchGrok Claims Elon Musk Is More Athletic Than LeBron James — and the World’s Greatest LoverWhat AI bubble? Nvidia's strong earnings signal there's more room to growAlphabet stock surges on Gemini 3 AI model optimismSunday Robotics emerges from stealth with launch of ‘Memo’ humanoid house chores robotRobotics Startup Physical Intelligence Valued at $5.6 Billion in New Funding - BloombergWaymo permitted areas expanded by California DMV - CBS Los AngelesWaymo enters 3 more cities: Minneapolis, New Orleans, and Tampa | TechCrunchMeta AI Releases Segment Anything Model 3 (SAM 3) for Promptable Concept Segmentation in Images and Videos - MarkTechPost[2511.16624] SAM 3D: 3Dfy Anything in Images[2511.13998] LoCoBench-Agent: An Interactive Benchmark for LLM Agents in Long-Context Software Engineering[2511.08544] LeJEPA: Provable and Scalable Self-Supervised Learning Without the Heuristics[2511.13720] Back to Basics: Let Denoising Generative Models DenoiseEurope is scaling back its landmark privacy and AI laws | The VergeFrom shortcuts to sabotage: natural emergent misalignment from reward hacking[2511.15304] Adversarial Poetry as a Universal Single-Turn Jailbreak Mechanism in Large Language ModelsDisrupting the first reported AI-orchestrated cyber espionage campaignOpenAI Locks Down San Francisco Offices Following Alleged Threat From Activist | WIREDWarner Music Group Settles AI Lawsuit With UdioSee Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

Our 225th episode with a summary and discussion of last week's big AI news! Recorded on 11/16/2025 Hosted by Andrey Kurenkov and co-hosted by Michelle Lee Feel free to email us your questions and feedback at contact@lastweekinai.com and/or hello@gladstone.ai Read out our text newsletter and comment on the podcast at https://lastweekin.ai/ In this episode: Timestamps: OpenAI says the brand-new GPT-5.1 is ‘warmer’ and has more ‘personality’ options | The VergeBaidu Unveils ERNIE 5.0 and a Series of AI Applications at Baidu World 2025, Ramps Up Global PushByteDance’s Volcano Engine debuts coding agent at $1.3 promo priceGoogle will let users call stores, browse products, and check out using AI | The VergeFei-Fei Li's World Labs speeds up the world model race with Marble, its first commercial product | TechCrunchOpenAI says it's fixed ChatGPT's em dash problem | TechCrunchAnthropic announces $50 billion data center plan | TechCrunchBaidu teases next-gen AI training, inference accelerators • The RegisterMeta chief AI scientist Yann LeCun plans to exit and launch own start-upAmazon Demands Perplexity Stop AI Tool From Making Purchases - BloombergAI PowerPoint-killer Gamma hits $2.1B valuation, $100M ARR, founder says | TechCrunchInception raises $50 million to build diffusion models for code and text | TechCrunchCoding assistant Cursor raises $2.3B 5 months after its previous round | TechCrunchChina's Baidu says it's running 250,000 robotaxi rides a week — same as Alphabet's WaymoDriverless Tech Firm Pony AI Raises $863 Million in HK ListingMoonshot's Kimi K2 Thinking emerges as leading open source AI[2510.26787] Remote Labor Index: Measuring AI Automation of Remote WorkOpenAI Researchers Train Weight Sparse Transformers to Expose Interpretable Circuits - MarkTechPostKimi Linear: An Expressive, Efficient Attention ArchitectureWatch Google DeepMind’s new AI agent learn to play video games | The VergearXiv Changes Rules After Getting Spammed With AI-Generated 'Research' PapersStability AI largely wins UK court battle against Getty Images over copyright and trademark | AP NewsCourt rules that OpenAI violated German copyright law; orders it to pay damages | TechCrunchMicrosoft's $15.2B UAE investment turns Gulf State into test case for US AI diplomacy | TechCrunchAn AI-Generated Country Song Is Topping A Billboard Chart, And That Should Infuriate Us All | Whiskey RiffXania Monet is the first AI-powered artist to debut on a Billboard airplay chart, but she likely won’t be the last | CNNElevenLabs’ new AI marketplace lets brands use famous voices for ads | The VergeSee Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

Our 224th episode with a summary and discussion of last week's big AI news! Recorded on 10/31/2025 Hosted by Andrey Kurenkov and co-hosted by Gavin Purcell (check out AI For Humans and AndThen!) Feel free to email us your questions and feedback at contact@lastweekinai.com and/or hello@gladstone.ai Read out our text newsletter and comment on the podcast at https://lastweekin.ai/ In this episode: Timestamps: Cursor 2.0 shifts to in-house AI with Composer model and parallel agentsAnthropic brings Claude Code to the web | TechCrunchMicrosoft's Mico is a 'Clippy' for the AI era | TechCrunchAnthropic’s Claude catches up to ChatGPT and Gemini with upgraded memory features | The VergeCanva launches its own design model, adds new AI features to the platform | TechCrunchElon Musk’s Grokipedia launches with AI-cloned pages from Wikipedia | The VergeOpenAI completed its for-profit restructuring — and struck a new deal with Microsoft | The VergeQualcomm announces AI chips to compete with AMD and NvidiaAmazon launches AI infrastructure project, to power Anthropic's Claude model | ReutersGoogle and Anthropic announce cloud deal worth tens of billionsGoogle partners with Ambani's Reliance to offer free AI Pro access to millions of Jio users in India | TechCrunchMiniMax Releases MiniMax M2: A Mini Open Model Built for Max Coding and Agentic Workflows at 8% Claude Sonnet Price and ~2x Faster - MarkTechPost[2510.25741] Scaling Latent Reasoning via Looped Language ModelsOpenAI’s gpt-oss-safeguard enables developers to build safer AI - Help Net Security[2510.15103] Continual Learning via Sparse Memory Finetuning[2510.18091] Accelerating Vision Transformers with Adaptive Patch Sizes[2510.18871] How Do LLMs Use Their Depth?AMD, Department of Energy announce $1 billion AI supercomputer partnership | The VergeUniversal partners with AI startup Udio after settling copyright suit | The VergeOpenAI loses bid to dismiss part of US authors' copyright lawsuit | ReutersSee Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.