English-language podcasts

The Daily AI Show

METR Inside Anthropic: 12x Gains From Unreleased Models

65 min

Listen through the publisher, then discuss the episode here or arrange an in-person Roundtable. Your city determines where to meet.

Roundtables are designed for groups of four, with a shared subject and room for everyone to speak. Come for discussion, with no dating expectation.

AI news today: METR's report from inside Anthropic finds unreleased models helping build the next generation and employees self-reporting 12x productivity gains, while the NASDAQ falls 5.5% in a day and Perplexity puts Teammates on its $200 Max plan. The conversation opens with hints of a new Qwen model and OpenAI's limited-access cyber model, then digs into METR's report on Anthropic: four internal-only models, internal models now used to build the next ones, roughly two years of progress in one year, and no obvious bottleneck. That leads to a market sell-off tied to data center resistance after the election, Nvidia's slide from about $240 to about $180, and what communities would need to get back before they welcome a data center. Also covered: Perplexity Teammates, an experiment where editing 0.016% of Qwen 3.5's weights lifted its math score from 7.2% to 42%, Meta's Muse avatar and planned device, Pino's World and an inbox-zero shop game, an Edmonton client where a receptionist's role grew into internal operations, the Dopplergeist audio geolocation tool, China flying five large aircraft without pilots, and the FAA retiring floppy disks by 2028. KEY POINTS DISCUSSED: 00:00:00 Qwen Model Hints and OpenAI's Cyber Model 00:01:48 METR Report on Anthropic's Internal Models 00:12:10 NASDAQ 5.5% Sell-Off and Nvidia's Slide 00:16:33 41 Data Center Ballot Votes, One Passed 00:22:40 Perplexity Teammates and the $200 Max Plan 00:25:46 Weight Edits Lift Qwen 3.5 Math Score 00:28:57 Meta's Muse Avatar and 'Furby' Device 00:32:04 Pino's World: 3eyez's Roaming AI Kiwi 00:34:03 Olivia Moore's Curio Shop Inbox Game 00:37:04 Edmonton Client: Receptionist Turned Operations Lead 00:52:46 Dopplergeist Audio Geolocation and China's Pilotless Planes 00:58:04 FAA Phasing Out Floppy Disks by 2028 The Daily AI Show Co Hosts: Beth Lyons, Andy Halliday, Gareth Hood, Karl Yeh

Episode information is provided by the publisher. Listening takes place on the publisher’s service.

Listen at the source ↗

Featured places to meet in Palo Alto, CA

Explore these places and arrange a conversation through ThinkMatch.

City guide for Palo Alto, CA

More from The Daily AI Show

The Daily AI Show

Anthropic Makes Billions, Gemini Neon, and Google's Project Genie

66 min

AI news from Anthropic's first day of trading, with Gemini Neon in Antigravity, Project Genie, Microsoft's MAI models, AI chips for China and Meta Muse. Anthropic priced its IPO at 83, opened near 111, and topped out above 114 before drifting back into the low 90s. Employees hold an estimated 232 billion dollars in stock, and Ashton Kutcher became a billionaire through his firm's early stake. Google stories included a hands-on account of Gemini Neon in Antigravity reviewing a whole project and then directing Claude Code as its product manager, a live Project Genie build of a South African safari, the Gemini Mac app's computer use, and Gemini for healthcare. Other topics were building an AI operations function from scratch at Verstela, gamified employee onboarding, Microsoft's Labs Canvas and MAI Code 2, House documents showing about 880,000 NVIDIA and AMD chips approved for export to China, and a Reuters report that Meta Muse surfaced private Signal messages through device permissions,

Discuss this episode

The Daily AI Show

Anthropic IPO Day, AI Music Awards, and AI Safety Warnings

56 min

AI news from the day Anthropic listed on Nasdaq, with AI safety warnings, model jailbreaks, robotics, AI music and new chips. Anthropic's Nasdaq debut was followed through the morning, and more than an hour after the opening bell its stock still had not started trading. Two safety stories were covered. A Center for AI Safety statement, signed by Nobel laureates and others, said Anthropic is not a safe AI company. A Government Accountability Office report found that ChatGPT, Claude, Gemini and Grok could be pushed past their guardrails into producing attack plans its experts judged viable. Other stories were Gemini decoding a 15th-century cipher, Costco panic buying tied to fears of AI cyberattacks, an ETH Zurich controller trained entirely in simulation to drive a John Deere front loader, the AI Music Awards and takedowns involving Suno and Sony, OpenAI's chip design work, and Gemini 4 Krypton, which was expected to drop that day. KEY POINTS DISCUSSED: 00:00:00 Anthropic IPO: Dario and

Discuss this episode

The Daily AI Show

GPT-6.2 Astra, AI Hack Fears, and Anthropic Voting Control

62 min

AI news on GPT-6.2 Astra's benchmarks and cost, a Java vulnerability, and NVIDIA's board leads into a talk about who controls AI companies. Artificial Analysis now ranks GPT-6.2 Astra Max second behind Fable 5.2 Max, and the hosts compare the two on cost ($4 versus $26 to run the full test suite), on what that gap means for subscribers, and on whether Astra 6.2 fits best as a general-purpose model. Speculation that recent outages are tied to AI-driven attacks leads into how downtime from patching differs from an actual breach, with the Java vulnerability and short Apple Pay and Chase interruptions as examples. The NVIDIA board's decision not to reinstate Jensen Huang, who holds 3.6% of NVIDIA's voting shares, opens a debate on founder control versus board accountability that extends to the single share giving Anthropic's founders 50.1% of shareholder votes. The hour ends on switching between Codex and Claude, testing models on personal use cases alongside benchmark results, and the ris

Discuss this episode

Community Poll

Cross-party cooperation is worthwhile even when it disappoints committed party supporters.

Share your view, then see how readers voted. No sign-in required. This poll is shared across stories.

Do you agree or disagree?

Choose one answer

Common Quandary

A Common Quandary to discuss

Another conversation to explore, separate from this news story.

An outsider becomes famous cooking another culture's food. Appreciation or taking over?

A chef becomes known for food from a culture outside their own after years of study with respected cooks. The restaurant is successful, while small businesses run by people from that culture struggle for attention.

When does appreciation become taking opportunity from the people who created a tradition?

Explore this quandary