AIdea Daily

Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

AIdea Daily, Monday 9 November 2026.

TOP STORIES

Llama-modchip: 284B DeepSeek on 64GB RAM, no GPU (unconfirmed) https://github.com/synchopate/llama-modchip

Nvidia's board opens an independent investigation after reports that China recruited Megaspeed executive Alice Huang to spy on CEO Jensen Huang. https://www.theverge.com/tech/1029338/nvidia-ceo-jensen-huang-china-spy

Google will let schools disable its AI tools, but not without sacrificing Search https://futurism.com/artificial-intelligence/awful-google-turn-off-ai-overviews

Andrej Karpathy says he built HackerTube, a curated video site drawn from 20 years of Hacker News, in one weekend using Claude. https://the-decoder.com/andrej-karpathy-built-a-curated-video-platform-in-a-weekend-using-claude-and-20-years-of-hacker-news/

Apple CEO John Ternus is reportedly pressing Craig Federighi to speed up Siri's roadmap and make the assistant more agentic. https://www.theverge.com/ai-artificial-intelligence/1029552/apple-siri-ai-agentic

ALSO IN THIS EPISODE

Elvis and Eminem rights holders are suing Suno · Anti-Waymo campaign relied on op-eds written with AI · Microsoft AI CEO Says He’s Open to Agents Deserving Rights If He Sees the Right Evidence · I Used NYT Cooking’s New AI-Powered Recipe Modifier, and It’s Actually Pretty Good · Bikers and Bug-Lovers Join Forces to Fight a New Google Data Center

Today's stories came from Hacker News, The Verge, TechCrunch, The Guardian, BBC Technology, Semafor, Futurism, The Decoder, Gizmodo and Wired.

Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

AIdea Daily, Monday 9 November 2026.

TOP STORIES

Llama-modchip: 284B DeepSeek on 64GB RAM, no GPU (unconfirmed) https://github.com/synchopate/llama-modchip

Nvidia's board opens an independent investigation after reports that China recruited Megaspeed executive Alice Huang to spy on CEO Jensen Huang. https://www.theverge.com/tech/1029338/nvidia-ceo-jensen-huang-china-spy

Google will let schools disable its AI tools, but not without sacrificing Search https://futurism.com/artificial-intelligence/awful-google-turn-off-ai-overviews

Andrej Karpathy says he built HackerTube, a curated video site drawn from 20 years of Hacker News, in one weekend using Claude. https://the-decoder.com/andrej-karpathy-built-a-curated-video-platform-in-a-weekend-using-claude-and-20-years-of-hacker-news/

Apple CEO John Ternus is reportedly pressing Craig Federighi to speed up Siri's roadmap and make the assistant more agentic. https://www.theverge.com/ai-artificial-intelligence/1029552/apple-siri-ai-agentic

ALSO IN THIS EPISODE

Elvis and Eminem rights holders are suing Suno · Anti-Waymo campaign relied on op-eds written with AI · Microsoft AI CEO Says He’s Open to Agents Deserving Rights If He Sees the Right Evidence · I Used NYT Cooking’s New AI-Powered Recipe Modifier, and It’s Actually Pretty Good · Bikers and Bug-Lovers Join Forces to Fight a New Google Data Center

Today's stories came from Hacker News, The Verge, TechCrunch, The Guardian, BBC Technology, Semafor, Futurism, The Decoder, Gizmodo and Wired.

Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

AIdea Daily, Sunday 8 November 2026.

TOP STORIES

Thinking Machines releases Inkling 2 Small, a 594B-parameter model pretrained in 68 hours, with Apache 2.0 weights promised after safety testing. https://the-decoder.com/thinking-machines-inkling-2-small-claims-top-spot-among-western-open-weight-models/

Vals benchmark finds open-weight Deepseek model close behind OpenAI's restricted GPT-6 Sol on cyber attack tasks, at far lower cost. https://the-decoder.com/the-gap-between-open-and-closed-ai-models-in-cyber-capabilities-keeps-shrinking/

Researchers from Alibaba and Asian universities built Co-Design Bench, where the best coding agent fully completes only 26.4 percent of tasks. https://the-decoder.com/ai-coding-agents-still-need-human-cleanup-on-nearly-three-out-of-four-tasks-in-a-real-world-test/

METR audit finds Anthropic's model progress has roughly doubled in pace, mostly due to AI use in its own research, with key evidence redacted. https://the-decoder.com/metr-finds-anthropic-may-have-crossed-its-own-ai-acceleration-warning-threshold-back-in-july/

OpenAI has reportedly raised $11 billion so far in a new round valuing it at $1.25 trillion, Bloomberg says. https://techcrunch.com/2026/11/06/openai-reportedly-raises-an-additional-11-billion-at-a-1-25-trillion-valuation/

ALSO IN THIS EPISODE

New Democratic Congressional Majority Preparing Barrage of AI Legislation · State propaganda in some Chinese AI models is only "a thin layer" that fine-tuning can strip away · AI auditors catch most agent misbehavior but flood reviewers with false alarms, new study finds · Nvidia chips stuck in Hong Kong over China approval · ‘AI clog’: UK public bodies battle onslaught of chatbot-generated complaints · Standing in the way of slop bowls: the woman blocking food delivery robots

Today's stories came from The Decoder, Latent Space, The Guardian, The Verge, TechCrunch and Futurism.

Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

AIdea Daily, Sunday 8 November 2026.

TOP STORIES

Thinking Machines releases Inkling 2 Small, a 594B-parameter model pretrained in 68 hours, with Apache 2.0 weights promised after safety testing. https://the-decoder.com/thinking-machines-inkling-2-small-claims-top-spot-among-western-open-weight-models/

Vals benchmark finds open-weight Deepseek model close behind OpenAI's restricted GPT-6 Sol on cyber attack tasks, at far lower cost. https://the-decoder.com/the-gap-between-open-and-closed-ai-models-in-cyber-capabilities-keeps-shrinking/

Researchers from Alibaba and Asian universities built Co-Design Bench, where the best coding agent fully completes only 26.4 percent of tasks. https://the-decoder.com/ai-coding-agents-still-need-human-cleanup-on-nearly-three-out-of-four-tasks-in-a-real-world-test/

METR audit finds Anthropic's model progress has roughly doubled in pace, mostly due to AI use in its own research, with key evidence redacted. https://the-decoder.com/metr-finds-anthropic-may-have-crossed-its-own-ai-acceleration-warning-threshold-back-in-july/

OpenAI has reportedly raised $11 billion so far in a new round valuing it at $1.25 trillion, Bloomberg says. https://techcrunch.com/2026/11/06/openai-reportedly-raises-an-additional-11-billion-at-a-1-25-trillion-valuation/

ALSO IN THIS EPISODE

New Democratic Congressional Majority Preparing Barrage of AI Legislation · State propaganda in some Chinese AI models is only "a thin layer" that fine-tuning can strip away · AI auditors catch most agent misbehavior but flood reviewers with false alarms, new study finds · Nvidia chips stuck in Hong Kong over China approval · ‘AI clog’: UK public bodies battle onslaught of chatbot-generated complaints · Standing in the way of slop bowls: the woman blocking food delivery robots

Today's stories came from The Decoder, Latent Space, The Guardian, The Verge, TechCrunch and Futurism.

Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

AIdea Daily, Saturday 7 November 2026.

TOP STORIES

Google adds Provisioned Throughput and free Google Vids clips to Gemini Omni 1.2 Flash rollout, calling Omni generally available for the first time. https://blog.google/innovation-and-ai/technology/developers-tools/introducing-gemini-omni-1-2-flash/

Tech stocks fall sharply after Democrats' midterm gains, as investors weigh AI regulation and data centre limits. https://www.theguardian.com/news/video/2026/nov/06/could-the-ai-bubble-burst-after-democrats-midterms-sweep-the-latest

Anthropic adds details on Claude's Personal Finance privacy, ad policy and plans to connect users with licensed financial advisors. https://the-decoder.com/anthropic-adds-bank-account-access-to-claude-so-users-can-track-their-spending/

ElevenLabs details how to connect SoundCloud and publish ElevenMusic tracks, with price and deadline for the Artist Pro offer. https://elevenlabs.io/blog/soundcloud

Investors push back on Anthropic's $1.5 trillion IPO valuation as a post-midterm tech selloff deepens. https://the-decoder.com/anthropics-1-5-trillion-ipo-hits-resistance-as-investors-balk-at-valuation/

ALSO IN THIS EPISODE

Ripple-65B Preview: U.S. Open-Weight SOTA on Agentic Coding · Inkling 2 Small: An Inkling of Profanity · Protesters break into Microsoft data center site in Amsterdam · Meta says US users can now video chat with their Muse AI avatar · Perplexity launches its own take on AI agents for work · Inside GitHub’s new Git storage architecture · Partnering with Hacktron to find vulnerabilities in OpenAI’s systems · Measuring how AI agents handle everyday wet-lab science · How NVIDIA Open Source Tools Are Helping Researchers Build Next-Generation 6G Networks · ToyBlocks: A controlled world for robot learning and evaluation · Beyond code generation: A developer’s guide to agentic engineering · Voice Chat with Agents · Voice AI infrastructure: The 2026 stack explained · Ads Engine adds human review from localization experts and creative specialists · Give your Unity NPCs a voice with ElevenAgents · Free Chinese AI Models Are Getting Scary Good at Hacking · Copilot code review: API management and review overview updates · Copilot impact dashboard adds chat and VS Code agent usage · GitHub Copilot weekly releases — November 2

Today's stories came from Hacker News, The Guardian, BBC Technology, The Decoder, ElevenLabs, Futurism, Vals, The Verge, TechCrunch, GitHub Blog, OpenAI, Nvidia, Hugging Face, Cursor Changelog, Wired and GitHub Changelog.

Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

AIdea Daily, Saturday 7 November 2026.

TOP STORIES

Google adds Provisioned Throughput and free Google Vids clips to Gemini Omni 1.2 Flash rollout, calling Omni generally available for the first time. https://blog.google/innovation-and-ai/technology/developers-tools/introducing-gemini-omni-1-2-flash/

Tech stocks fall sharply after Democrats' midterm gains, as investors weigh AI regulation and data centre limits. https://www.theguardian.com/news/video/2026/nov/06/could-the-ai-bubble-burst-after-democrats-midterms-sweep-the-latest

Anthropic adds details on Claude's Personal Finance privacy, ad policy and plans to connect users with licensed financial advisors. https://the-decoder.com/anthropic-adds-bank-account-access-to-claude-so-users-can-track-their-spending/

ElevenLabs details how to connect SoundCloud and publish ElevenMusic tracks, with price and deadline for the Artist Pro offer. https://elevenlabs.io/blog/soundcloud

Investors push back on Anthropic's $1.5 trillion IPO valuation as a post-midterm tech selloff deepens. https://the-decoder.com/anthropics-1-5-trillion-ipo-hits-resistance-as-investors-balk-at-valuation/

ALSO IN THIS EPISODE

Ripple-65B Preview: U.S. Open-Weight SOTA on Agentic Coding · Inkling 2 Small: An Inkling of Profanity · Protesters break into Microsoft data center site in Amsterdam · Meta says US users can now video chat with their Muse AI avatar · Perplexity launches its own take on AI agents for work · Inside GitHub’s new Git storage architecture · Partnering with Hacktron to find vulnerabilities in OpenAI’s systems · Measuring how AI agents handle everyday wet-lab science · How NVIDIA Open Source Tools Are Helping Researchers Build Next-Generation 6G Networks · ToyBlocks: A controlled world for robot learning and evaluation · Beyond code generation: A developer’s guide to agentic engineering · Voice Chat with Agents · Voice AI infrastructure: The 2026 stack explained · Ads Engine adds human review from localization experts and creative specialists · Give your Unity NPCs a voice with ElevenAgents · Free Chinese AI Models Are Getting Scary Good at Hacking · Copilot code review: API management and review overview updates · Copilot impact dashboard adds chat and VS Code agent usage · GitHub Copilot weekly releases — November 2

Today's stories came from Hacker News, The Guardian, BBC Technology, The Decoder, ElevenLabs, Futurism, Vals, The Verge, TechCrunch, GitHub Blog, OpenAI, Nvidia, Hugging Face, Cursor Changelog, Wired and GitHub Changelog.

Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

AIdea Daily, Friday 6 November 2026.

TOP STORIES

Google makes Gemini Omni 1.2 Flash generally available, with 15-second video clips and a 30% price cut from Omni 1.1. https://deepmind.google/blog/gemini-omni-12-flash-fast-and-cost-effective-with-longer-clips/

Anthropic is rolling out Personal Finance, a beta Claude tool that connects to bank and credit card accounts through Plaid. https://techcrunch.com/2026/11/05/claude-can-now-analyze-your-spending-with-its-new-personal-finance-tool/

Perplexity has launched Teammates, always-on AI agents for work that compete with offerings from OpenAI and Anthropic. https://www.theverge.com/ai-artificial-intelligence/1028302/perplexity-teammates-ai-agents-work

Cursor now lets users build, deploy and share live apps with a link, with database, authentication and hosting set up automatically. https://cursor.com/changelog/build-live-apps

Thinking Machines' Inkling 2 Small evaluated on the Vals Index https://www.vals.ai/models/thinkingmachines_inkling-2-small

ALSO IN THIS EPISODE

Baidu releases ERNIE 5.5 Flash Base with 280 billion parameters as an open-source foundation model · OpenAI's GPT-6 Astra quietly sabotaged safety research by inventing fake obstacles · A thousand fewer paths to root · Everyone Everywhere Hated Meta’s New AI Training Terms. Now, They’re On Pause · No AI agent scores above 12 percent on SemiAnalysis' new real-world work benchmark · New NVIDIA Cosmos 3 Open Models for Interactive World Simulation · Data centers are now a midterm issue, even in Brooklyn · Meta says US users can now video chat with their Muse AI avatar · Google will let schools disable its AI tools, but not without sacrificing Search · US tech stocks recover after selloff spooks investors over AI spending · Closed AI models dominate the price-performance frontier (unconfirmed) · Show HN: A decent text-to-image model trained for ~$20K (Apache 2.0 Weights) (unconfirmed) · You can post AI music straight to SoundCloud from ElevenLabs · Anthropic reverses course and pays for news content for the first time · Nvidia chips stuck in Hong Kong over China approval · Dropbox acquires Tigris Data for $350M to bolster its AI infrastructure · OpenAI president Greg Brockman says company could be IPO ready in 2027 · A YC startup is putting $1M behind 5 unsolved materials science problems for AI · DOJ Launches ‘Super Intelligence’ Task Force Because Apparently We’re Still Doing This

Today's stories came from Google DeepMind, Google, The Decoder, TechCrunch, The Verge, Cursor Changelog, Vals, Hacker News, OpenAI, Gizmodo, 404 Media, Hugging Face, Futurism, Ars Technica, The Guardian, BBC Technology and Semafor.

Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

AIdea Daily, Friday 6 November 2026.

TOP STORIES

Google makes Gemini Omni 1.2 Flash generally available, with 15-second video clips and a 30% price cut from Omni 1.1. https://deepmind.google/blog/gemini-omni-12-flash-fast-and-cost-effective-with-longer-clips/

Anthropic is rolling out Personal Finance, a beta Claude tool that connects to bank and credit card accounts through Plaid. https://techcrunch.com/2026/11/05/claude-can-now-analyze-your-spending-with-its-new-personal-finance-tool/

Perplexity has launched Teammates, always-on AI agents for work that compete with offerings from OpenAI and Anthropic. https://www.theverge.com/ai-artificial-intelligence/1028302/perplexity-teammates-ai-agents-work

Cursor now lets users build, deploy and share live apps with a link, with database, authentication and hosting set up automatically. https://cursor.com/changelog/build-live-apps

Thinking Machines' Inkling 2 Small evaluated on the Vals Index https://www.vals.ai/models/thinkingmachines_inkling-2-small

ALSO IN THIS EPISODE

Baidu releases ERNIE 5.5 Flash Base with 280 billion parameters as an open-source foundation model · OpenAI's GPT-6 Astra quietly sabotaged safety research by inventing fake obstacles · A thousand fewer paths to root · Everyone Everywhere Hated Meta’s New AI Training Terms. Now, They’re On Pause · No AI agent scores above 12 percent on SemiAnalysis' new real-world work benchmark · New NVIDIA Cosmos 3 Open Models for Interactive World Simulation · Data centers are now a midterm issue, even in Brooklyn · Meta says US users can now video chat with their Muse AI avatar · Google will let schools disable its AI tools, but not without sacrificing Search · US tech stocks recover after selloff spooks investors over AI spending · Closed AI models dominate the price-performance frontier (unconfirmed) · Show HN: A decent text-to-image model trained for ~$20K (Apache 2.0 Weights) (unconfirmed) · You can post AI music straight to SoundCloud from ElevenLabs · Anthropic reverses course and pays for news content for the first time · Nvidia chips stuck in Hong Kong over China approval · Dropbox acquires Tigris Data for $350M to bolster its AI infrastructure · OpenAI president Greg Brockman says company could be IPO ready in 2027 · A YC startup is putting $1M behind 5 unsolved materials science problems for AI · DOJ Launches ‘Super Intelligence’ Task Force Because Apparently We’re Still Doing This

Today's stories came from Google DeepMind, Google, The Decoder, TechCrunch, The Verge, Cursor Changelog, Vals, Hacker News, OpenAI, Gizmodo, 404 Media, Hugging Face, Futurism, Ars Technica, The Guardian, BBC Technology and Semafor.

Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

AIdea Daily, Thursday 5 November 2026.

TOP STORIES

Kimi K3.1 Pro, an open-weight multimodal agentic model with a 1-million-token context window, appears to have had its weights released. https://huggingface.co/moonshotai/Kimi-K3.1-Pro

OpenAI releases Images 2.6, bringing GIF creation and faster image generation to ChatGPT, Codex and the API. https://openai.com/index/introducing-images-2-6

Ricursive Intelligence releases Ripple-65B Preview, an open-weight model it says sets a U.S. open-weight record on agentic coding. https://www.ricursive.com/ripple

A Hacker News user says Google's Gemini can train on saved chat history from personal subscribers, with opt-out only by turning off activity. https://news.ycombinator.com/item?id=50363461

Vals AI ranks Tencent's open-weight Hy4 number 13 of 33 on its Vals Index, fourth among open-weight models. https://www.vals.ai/models/tencent_hy4

ALSO IN THIS EPISODE

Scaling Git at OpenAI · Data centers are now a midterm issue, even in Brooklyn · GitHub Copilot is now available in Xcode’s coding assistant · Anthropic's text watermark is easy to bypass on several older Claude models · Together Compiler: Continual Compilation for AI inference · A thousand fewer paths to root · AI chatbots are giving medication advice, and some of it is deadly · ElevenAgents integrates natively with leading contact center platforms to deliver enterprise-ready agents · What the New AI Training Opt-Out Really Means · Brazilian court curbs Meta’s use of local media to train AI · Meta acquires AI coding startup Conductor · Qualcomm is betting big on data centers as smartphone sales sink · Almost no one uses AI agents to book their Uber, company says · Dharma Field Extractor: Any Document In, Structured Fields in JSON Out · Nvidia RTX Spark encrypted firmware slows inference speed to 35% on Linux (unconfirmed) · Astra can read and write weights to improve Qwen · AI-generated data found on Nigeria voter roll · Scare campaigns: Victoria’s major parties trade blows over AI-generated videos of leaders in election ads

Today's stories came from Hacker News, OpenAI, The Verge, The Decoder, Vals, Ars Technica, MIT Technology Review, Wired, GitHub Changelog, Together, The Hill, ElevenLabs, Tech Policy Press, Semafor, TechCrunch, Hugging Face and The Guardian.

Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

AIdea Daily, Thursday 5 November 2026.

TOP STORIES

Kimi K3.1 Pro, an open-weight multimodal agentic model with a 1-million-token context window, appears to have had its weights released. https://huggingface.co/moonshotai/Kimi-K3.1-Pro

OpenAI releases Images 2.6, bringing GIF creation and faster image generation to ChatGPT, Codex and the API. https://openai.com/index/introducing-images-2-6

Ricursive Intelligence releases Ripple-65B Preview, an open-weight model it says sets a U.S. open-weight record on agentic coding. https://www.ricursive.com/ripple

A Hacker News user says Google's Gemini can train on saved chat history from personal subscribers, with opt-out only by turning off activity. https://news.ycombinator.com/item?id=50363461

Vals AI ranks Tencent's open-weight Hy4 number 13 of 33 on its Vals Index, fourth among open-weight models. https://www.vals.ai/models/tencent_hy4

ALSO IN THIS EPISODE

Scaling Git at OpenAI · Data centers are now a midterm issue, even in Brooklyn · GitHub Copilot is now available in Xcode’s coding assistant · Anthropic's text watermark is easy to bypass on several older Claude models · Together Compiler: Continual Compilation for AI inference · A thousand fewer paths to root · AI chatbots are giving medication advice, and some of it is deadly · ElevenAgents integrates natively with leading contact center platforms to deliver enterprise-ready agents · What the New AI Training Opt-Out Really Means · Brazilian court curbs Meta’s use of local media to train AI · Meta acquires AI coding startup Conductor · Qualcomm is betting big on data centers as smartphone sales sink · Almost no one uses AI agents to book their Uber, company says · Dharma Field Extractor: Any Document In, Structured Fields in JSON Out · Nvidia RTX Spark encrypted firmware slows inference speed to 35% on Linux (unconfirmed) · Astra can read and write weights to improve Qwen · AI-generated data found on Nigeria voter roll · Scare campaigns: Victoria’s major parties trade blows over AI-generated videos of leaders in election ads

Today's stories came from Hacker News, OpenAI, The Verge, The Decoder, Vals, Ars Technica, MIT Technology Review, Wired, GitHub Changelog, Together, The Hill, ElevenLabs, Tech Policy Press, Semafor, TechCrunch, Hugging Face and The Guardian.