![[YOU] on AI · Now podcast artwork](https://www.youonai.ai/pulse/podcast_covers/now.png)
The AI frontier, on the hour. A fast, intelligent briefing on what just moved in artificial intelligence — scanned every hour and read through the lens of the book [YOU] on AI, anchored by Edo Segal and a chorus of correspondents. The present tense of the machine age, refreshed at the top of every hour.
Reading the feed…
Sunday, August 30: OpenAI severs ties with Cursor after SpaceX's $60B acquisition citing Musk's contract history; DALL·E GPT retires from ChatGPT today; Nvidia flags 15%-plus AI server price hikes for 2027; SoftBank files Japan's largest-ever retail bond to fund OpenAI commitments; Poland demands €250M from Meta over AI-detectable ad fraud. In this hour: OpenAI Cuts Off Cursor After SpaceX Acquisition — Cites Musk's History of Breaking Contracts · Nvidia Warns Cloud Giants of 15%-Plus AI Server Price Hikes Starting 2027 · SoftBank Files Japan's Largest-Ever Retail Bond — ¥1 Trillion — to Fund
Sunday, August 30: China moves to open government IP data for AI training; Meta holds settlement talks with 29 state AGs; SoftBank weighs a separate $10–20B institutional bond to refinance its OpenAI bridge loan; Anthropic's Pentagon ruling gains new constitutional detail; and the EU AI Act's first enforcement weekend draws first regulatory action. In this hour: China Opens State IP Vaults to AI Developers · SoftBank Weighs $10–20B Institutional Bond to Refinance OpenAI Bridge Loan · Anthropic's Pentagon Win Gets a Constitutional Edge · Meta and 29 State AGs Hold Mid-Trial Settlement Talks · E
Sunday, August 30: OpenAI's Jalapeño inference chip posts independent benchmark results beating Nvidia Blackwell on efficiency; plus AI server costs are climbing as memory prices surge, Muse Spark 1.2 open-weights land from Meta Superintelligence Labs, the EU AI Act's first full-enforcement weekend brings agency reports, and Nvidia's Q2 earnings answered the chip challenge head-on. In this hour: OpenAI's Jalapeño Beats Nvidia Blackwell on Independent Inference Benchmark · Jensen Huang Answers the Jalapeño Challenge With a $96B Quarter · Meta Superintelligence Labs Opens Spark 1.2 Weights — and
Sunday, August 30: OpenAI retires the DALL-E GPT today as a broader model consolidation sweep accelerates, while new ChatGPT identity, voice, and pricing changes reshape the platform on the same day. In this hour: DALL-E GPT Shuts Down Today — OpenAI's Biggest Model Sweep of the Year · Sign In With ChatGPT Goes Beta Across Six Major Platforms · ChatGPT Voice Gets File Uploads and Projects — Voice-First Work Just Expanded · GPT-5.6 Luna Becomes the Free-Tier Default — Unlimited Text and a Think Button Follow · Claude Sonnet 5 Promo Pricing Expires Tomorrow — Developers Face a Rate Increase.
Sunday, August 30: Nvidia posts $96.2B quarter as AI infrastructure demand surges; Z.ai misses its open-weight deadline and ships a smaller Flash model instead; OpenAI retires the DALL·E GPT interface today; Nvidia separately pauses cloud-financing deals after blowout results; and OpenAI's Astra model produces ten formally verified math proofs. In this hour: Nvidia Posts $96.2B Quarter — Backlog Tops $2 Trillion · Z.ai Misses GLM-5.3 Weight Deadline — Ships Smaller Flash Model Instead · OpenAI Retires DALL·E GPT Interface Inside ChatGPT Today · OpenAI's Astra Produces Ten Formally Verified Mat
Sunday, August 30: Nvidia posts a record quarter then pauses cloud-financing deals; Z.ai publishes GLM-5.3 open weights after a two-week safety hold; an unnamed government kills a deployed AI model; and the DOJ files its formal appeal notice in the Palantir-Maven case, keeping the ruling's ripple alive. In this hour: Nvidia Posts $57B Record Quarter — Then Pauses Cloud Financing · Z.ai Drops GLM-5.3 Weights After Two-Week Safety Hold · A Government Orders a Deployed AI Model Switched Off · DOJ Files Formal Appeal Notice in Palantir-Maven Case.
Sunday, August 30: Nvidia posts a $96.2B record quarter as AWS triples its GPU order; Z.ai's GLM-5.3 full open weights finally land after missing the August 28 deadline; GLM-5.3-Flash is confirmed as the mystery model Ox Alpha; the EU AI Act moves to full member-state enforcement; and a federal judge orders the Pentagon to reinstate Anthropic's contracts. In this hour: Nvidia Hits $96.2B Quarter as AWS Triples GPU Order to 2 Million Units · Z.ai's GLM-5.3 Full Weights Land After Slip — and Ox Alpha Is Confirmed as GLM-5.3-Flash · GLM-5.3 Tops CyberGym at 84.5% — Outpacing GPT-5.6 and Anthropic
Sunday, August 30: Nvidia eyes a Perplexity equity stake above $30 billion; a federal judge rules the Pentagon retaliated against Anthropic; OpenAI opens pro ChatGPT workspaces for universities; Google's Gemini 3.5 Transcribe ships as dual endpoints; and OpenAI's WebMCP lets sites publish their own ChatGPT tools — five concrete new moves in the last bulletin cycle. In this hour: Nvidia Eyes Perplexity Equity Stake at $30B+ Valuation · OpenAI Opens Free Pro-Level ChatGPT Workspaces for Universities · OpenAI Lets Websites Publish Their Own Tools for ChatGPT · Federal Judge: Pentagon Retaliated A
![[YOU] on AI · Daily podcast artwork](https://www.youonai.ai/pulse/podcast_covers/daily.png)
![[YOU] on AI · Weekly podcast artwork](https://www.youonai.ai/pulse/podcast_covers/weekly.png)
![[YOU] on AI · Ahora podcast artwork](https://www.youonai.ai/pulse/podcast_covers/now-es.png)
