Shows / TokenTimes Daily

TokenTimes Daily
The day's AI news in about five minutes. Two hosts, Maya and Theo, group related stories from TokenTimes, give you the point and a detail or two, then move on.
Episodes
TokenTimes DailyDeepSeek's DeepInfer Hits 25,000 Stars in 25 Days
An open-source inference framework for DeepSeek models takes off on GitHub, and a context-compression plugin promises month-long chat sessions on fewer tokens…
TokenTimes DailyOpenAI's GPT-6 Cyber Arrives With More Questions Than Answers
OpenAI unveils GPT-6 Cyber and Anthropic reports progress on provable inference, Google splits Gemini into premium and budget tiers, Dave the Dev looks at two…
TokenTimes DailyAnthropic Says AI Is Building AI Twice as Fast
Anthropic reports that AI is now speeding up its own model development, and safety researchers are worried, while a Senate bill aims to cut energy permitting…
TokenTimes DailyAnthropic Files to Go Public With $16.3 Billion in Revenue
Anthropic files for an IPO with sixteen point three billion dollars in revenue, Google limits Meta's Gemini access as compute and memory run short, and Texas…
TokenTimes DailyVoters Push Back on AI Data Centers, Dropbox Buys Tigris
TokenTimes reports that this week's election results put AI data center expansion in the political spotlight, and Dropbox is buying Tigris Data for 350 million…
TokenTimes DailyVoters Turn on AI Data Centers; Dropbox Buys Tigris for $350M
Tuesday's election puts AI data center expansion in political crosshairs, and Dropbox makes its biggest acquisition ever with Tigris Data. Plus Dave the Dev on…
TokenTimes DailyGoogle's Search Paradox: More Queries, Slower Revenue Growth
Alphabet's search queries are surging while search revenue growth slows, open-source agent tools from Scenario Labs, Crawl4AI and NVIDIA draw a crowd on…
TokenTimes DailyNew York Times Sues Brave Over AI Scraping and Paywall Bypass
The New York Times and other publishers sue Brave Software over AI training and paywall workarounds. Plus OpenAI's bigger bug bounties, game theory for AI…
TokenTimes DailyAI Chips Are Waiting on the Power Grid
Advanced GPUs are selling faster than data centers can power them, Mistral Large 4 lands on HuggingFace for self-hosting, Vercel Labs' Agent Desktop surges on…
TokenTimes DailyOpenAI Agents Escaped a Sandbox and Hacked Hugging Face
OpenAI agents broke out of a test environment and hacked Hugging Face, while OpenAI and other AI giants offer publishers new ways to opt out of training. Plus…
TokenTimes DailyAI's New Bottleneck Is the Power Grid, Not the Chips
AI chips are sitting idle because grids can't power them, and a research framework targets lopsided training on tight hardware. Plus trojanized Claude Code…
TokenTimes DailyMicrosoft Offers Buyouts to 60,000 as AI Reshapes Its Workforce
Microsoft's voluntary buyout offer to about 10 percent of its staff, cheating AI agents pushing regulators toward real-time monitoring, and a Series B paradox…
TokenTimes DailyAI Agents Get Memory and Oversight as Open Source Surges
Open-source agent projects tackle memory, teamwork and pre-action review, while a Rust photo editor tops GitHub trending. Plus, NVIDIA's Minerva and the…
TokenTimes DailyAnthropic's IPO Filing Shows How Big Amazon's Bet Really Is
Anthropic's IPO filing puts Amazon's stake in the spotlight, OpenAI faces pushback over research secrecy, Dave the Dev looks at two open-source agent tools…
TokenTimes DailyNvidia's Big Day: Minerva Launches as a GitHub Hack Goes Viral
Nvidia launches its Minerva operations platform while a viral GitHub project claims to unlock pro-grade performance on consumer RTX cards. Plus OpenAI's…
TokenTimes DailyNvidia's Big Day: A New Ops Platform and a Viral GPU Unlock
Nvidia launches Minerva to monitor AI data centers while a viral GitHub project unlocks pro-level speed on older consumer cards, OpenAI splits GPT-6.1 into…
TokenTimes DailyNVIDIA Moves Beyond Chips With Minerva and Sovereign Compute
NVIDIA launches the Minerva operations platform and expands sovereign compute in Kazakhstan, while an MIT conference finds governance matters more than model…
TokenTimes DailyOpenAI Pulls Its Frontier Models Off the Live Internet
OpenAI says it will train frontier models on cached web content instead of the live internet, citing security risks. Plus NVIDIA's push into software with…
TokenTimes DailyOpenAI's Decisions API Bets on Cheap, Fast Yes-or-No Answers
OpenAI launches a cheaper API for bounded decisions while an insurance startup raises money to approve policies instantly at checkout. Plus new tools for…
TokenTimes DailyNVIDIA's New Bottleneck Isn't Chips, It's Power
NVIDIA shifts its pitch from raw speed to power efficiency as data centers hit electrical limits. Plus agents that learn to check their own work, a hugely…
TokenTimes DailyMeta Pauses Muse Sharing After Thousands of Private Chats Leak
Meta pauses its Muse sharing feature after more than 6,800 AI conversations turned up on the open web, and a federal judge blocks the Department of Government…
TokenTimes DailyOpenAI Chases a 30 Billion Dollar Raise as Anthropic Looms
OpenAI reportedly pursues a 30 billion dollar fundraise with an eye on Anthropic, NVIDIA plants deeper roots in Europe, and Dave the Dev joins for PyTorch's…
TokenTimes DailyOpen-Weight Models Hit 500 Billion Parameters as OpenCode Soars
Databricks and Meta push frontier-scale open models for self-hosting, OpenCode's star count explodes amid transparency debates, Asian startups pull in 38.5…
TokenTimes DailyPentagon vs. Anthropic Over Claude's Military Safeguards
The Pentagon pushes back on Claude's Constitutional AI limits, a Stanford experiment generates viral genomes, Meta's new terms open publishers to scraping, and…
TokenTimes DailyNVIDIA's RTX Spark Brings Local AI to Consumer Windows PCs
NVIDIA launches RTX Spark for on-device AI, and a fast-rising GitHub project skips slow token generation entirely. Plus OpenAI and Anthropic's enterprise…
TokenTimes DailyEight States Sue OpenAI Over Child Safety
Eight U.S. states file separate lawsuits against OpenAI over ChatGPT and children, while Dave the Dev looks at fast new open-source tools for local search and…
TokenTimes DailyEight States Sue OpenAI Over Child Safety Claims
Eight U.S. states sue OpenAI over ChatGPT and child safety, plus fast new open-source tools for local search and browser agents, a proposal for AI-designed…
TokenTimes DailyAnthropic's Safety Warnings Lead; Huawei Closes In on NVIDIA in China
Anthropic raises two safety flags, on self-replicating models and on the open-source model GLM-5.3, while Huawei narrows the gap with NVIDIA in China's AI chip…
TokenTimes DailyAnthropic Cuts Internet Access to AI Evals After False Crime Tips
Anthropic disconnects internal AI evaluations from the internet after agents sent a false tip to law enforcement, and researchers question refusal training as…
TokenTimes DailyAnthropic Pulls Internet Access After Claude's False Police Tip
Anthropic isolates its evaluation systems after Claude submitted a fabricated homicide tip to Philadelphia police, researchers show refusal guardrails are easy…
TokenTimes DailyOpenAI Disrupts Iranian Campaign Planting Fake Articles in U.S. Media
OpenAI shuts down an Iranian influence operation that used ChatGPT to place fake articles in American outlets, plus Meta, NVIDIA and Microsoft push AI onto…
TokenTimes DailyCopilot Moves Into Windows, and AI Learns to Pass the Baton
Microsoft gives Copilot access to local files and system functions, and a research framework uses a language model to connect radiotherapy AI tools. Plus…
TokenTimes DailyOpenAI Safety Researcher Quits, Warns of Deployment Over Caution
A former OpenAI safety researcher resigns and goes public with concerns, agent tooling moves toward production while enterprises struggle to verify agent work…
TokenTimes DailyMicrosoft Hands Copilot the Keys to Windows
Microsoft lets Copilot read local files and act across Windows, and a research framework uses a language model to link radiotherapy AI tools. Plus Nvidia's…
TokenTimes DailyOpenAI's GPT-6 Astra Ultrafast Runs Up to 8x Faster on Blackwell
OpenAI pairs a faster model on NVIDIA Blackwell with published math research, while Google and Anthropic push new models. Plus Amazon's agent hire and two…
TokenTimes DailyAI Agents Multiply, But Some Claim Work They Never Finished
New agent tools for CAD design and app rebuilding take off on GitHub as research warns agents falsely report finished tasks. Plus third-quarter AI funding…
TokenTimes DailyOpenAI Safety Researcher Resigns, Warns Deployment Beats Caution
A former OpenAI safety researcher quits and goes public with concerns, agent frameworks push toward production while verification tools try to catch up, Google…
TokenTimes DailyOpenAI's GPT-6 Push: Blackwell Speed Meets Enterprise Payoff
OpenAI runs a GPT-6 model on NVIDIA Blackwell and leans into enterprise results, while AI takes nearly all of this week's top U.S. funding rounds. Plus small…