<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0"><channel>
<title>AIPulse (en)</title>
<link>https://aipulseai.com/</link>
<description>AIPulse: hourly AI news from 47 real sources, deduplicated and scored, in English, Chinese and Spanish. Model leaderboard, price tracking and daily digest.</description>
<language>en</language>
<lastBuildDate>Mon, 09 Nov 2026 18:07:18 GMT</lastBuildDate>
<generator>AIPulse static generator</generator>
<item><title>Meta releases Muse Ember, a new lightweight model for agent execution</title><link>https://techcrunch.com/2026/11/09/meta-releases-muse-ember-a-new-lightweight-model-for-agent-execution/</link><guid isPermaLink="true">https://techcrunch.com/2026/11/09/meta-releases-muse-ember-a-new-lightweight-model-for-agent-execution/</guid><description>At 10 cents per million input tokens, Ember is significantly cheaper than frontier competitors like Haiku and Luna. What it means: Fresh from TechCrunch AI — read the original for full context.</description><pubDate>Mon, 09 Nov 2026 18:00:00 GMT</pubDate></item>
<item><title>James Gunn is out as DC movie leader</title><link>https://www.theverge.com/entertainment/1030022/james-gunn-peter-safran-dc-studios-out-leaving</link><guid isPermaLink="true">https://www.theverge.com/entertainment/1030022/james-gunn-peter-safran-dc-studios-out-leaving</guid><description>Soon, James Gunn and Peter Safran will no longer head up Skydance's DC Studios. After taking the helm of DC Studios as co-CEOs in 2022, Gunn and Safran are being replaced by former Marvel SVP of Production and Development Jeremy Latcham, Deadline reports. However, Gunn and Safran will &quot;continue to produce on various DC projects&quot; [&amp;#8230;] What it means: Fresh from The Verge — read the original for full context.</description><pubDate>Mon, 09 Nov 2026 17:39:57 GMT</pubDate></item>
<item><title>The future of AI hinges on… Jensen Huang’s sex life?</title><link>https://www.theverge.com/ai-artificial-intelligence/1029990/jensen-huang-the-spy-who-loved-me</link><guid isPermaLink="true">https://www.theverge.com/ai-artificial-intelligence/1029990/jensen-huang-the-spy-who-loved-me</guid><description>I would prefer that these people's sex lives were none of my business but I guess we have to talk about Jensen Huang and the National Security Incident. Who could have guessed that the guy who was out here signing titties might not have great judgment about how to behave around women? Alice Huang - [&amp;#8230;] What it means: Fresh from The Verge — read the original for full context.</description><pubDate>Mon, 09 Nov 2026 17:36:08 GMT</pubDate></item>
<item><title>Google Unveils TPU 8i Max and TPU 8tx: A Dual-Chip TPU and a Confidential TPU</title><link>https://www.marktechpost.com/2026/11/09/google-tpu-8i-max-tpu-8tx/</link><guid isPermaLink="true">https://www.marktechpost.com/2026/11/09/google-tpu-8i-max-tpu-8tx/</guid><description>Google added TPU 8i Max and TPU 8tx to its eighth-generation TPU lineup. TPU 8i Max packs 2 TPU 8i chips per package for long-context agents. TPU 8tx adds hardware confidential inference. Bloomberg reports 2027 availability. The post Google Unveils TPU 8i Max and TPU 8tx: A Dual-Chip TPU and a Confidential TPU appeared first on MarkTechPost . What it means: Fresh from MarkTechPost — read the original for full context.</description><pubDate>Mon, 09 Nov 2026 17:09:54 GMT</pubDate></item>
<item><title>Anthropic joins AI's natural gas frenzy, buys $3.4B worth of turbines</title><link>https://techcrunch.com/2026/11/09/anthropic-joins-ais-natural-gas-frenzy-buys-3-4b-worth-o</link><guid isPermaLink="true">https://techcrunch.com/2026/11/09/anthropic-joins-ais-natural-gas-frenzy-buys-3-4b-worth-o</guid><description>Anthropic bought $3.4 billion worth of natural gas turbines to power its AI data centers, joining the industry's rush to secure its own electricity supply. What it means: AI labs are now energy companies in disguise — whoever controls power controls training capacity.</description><pubDate>Mon, 09 Nov 2026 16:56:00 GMT</pubDate></item>
<item><title>Samsung mulls Galaxy AI usage limits and subscription tiers – what we know</title><link>https://www.zdnet.com/innovation/samsung-galaxy-ai-paid-subscription/</link><guid isPermaLink="true">https://www.zdnet.com/innovation/samsung-galaxy-ai-paid-subscription/</guid><description>Galaxy device owners who are heavy AI users would likely start paying for some features they’ve long enjoyed for free. What it means: Fresh from ZDNet AI — read the original for full context.</description><pubDate>Mon, 09 Nov 2026 16:55:56 GMT</pubDate></item>
<item><title>Baidu’s ERNIE Evolve Automated Research System Beats NanoChat and NanoGPT Records</title><link>https://www.marktechpost.com/2026/11/09/baidu-ernie-evolve-automated-research-system/</link><guid isPermaLink="true">https://www.marktechpost.com/2026/11/09/baidu-ernie-evolve-automated-research-system/</guid><description>Baidu’s ERNIE Evolve is an automated research system that edits code, runs experiments and keeps only verified gains. It beat public bests on NanoChat and NanoGPT Speedrun and sped up ERNIE-5.5-Flash inference. The post Baidu’s ERNIE Evolve Automated Research System Beats NanoChat and NanoGPT Records appeared first on MarkTechPost . What it means: Fresh from MarkTechPost — read the original for full context.</description><pubDate>Mon, 09 Nov 2026 16:31:59 GMT</pubDate></item>
<item><title>Trump's 60 Minutes interview touched on AI safety, threat to arrest an AI exec</title><link>https://www.theverge.com/ai-artificial-intelligence/1029802/trump-60-minutes-ai</link><guid isPermaLink="true">https://www.theverge.com/ai-artificial-intelligence/1029802/trump-60-minutes-ai</guid><description>In a 60 Minutes interview, President Trump dismissed AI executives' safety warnings — “the risk is not nearly as real as they talk” — and said the Justice Department is managing the risk, after earlier raising the prospect of arresting an AI executive. What it means: Regulation by vibes: the White House is signaling it will not slow AI down, which is itself the biggest policy signal of the year.</description><pubDate>Mon, 09 Nov 2026 15:50:00 GMT</pubDate></item>
<item><title>Tencent's AI chief: AI bioweapons are coming, but closing off models isn't the answer</title><link>https://the-decoder.com/tencents-ai-chief-says-ai-bioweapons-are-coming-but-closing-off-models-isnt-the-answer/</link><guid isPermaLink="true">https://the-decoder.com/tencents-ai-chief-says-ai-bioweapons-are-coming-but-closing-off-models-isnt-the-answer/</guid><description>Tencent chief AI scientist Yao Shunyu warns AI will soon be able to develop bioweapons, and argues for a middle ground: closed models block too aggressively, while open models' safety filters can be stripped — so Tencent filters dangerous knowledge out of its upcoming Hy4 model's training data. What it means: The safety debate is shifting from “open vs closed” to “what never goes into training data at all”.</description><pubDate>Mon, 09 Nov 2026 15:44:00 GMT</pubDate></item>
<item><title>ChatGPT now takes over 81% of global AI chatbot traffic</title><link>https://www.ithome.com/1/024/542.htm</link><guid isPermaLink="true">https://www.ithome.com/1/024/542.htm</guid><description>ChatGPT now accounts for over 81% of global AI chatbot traffic, according to a Graphite study — no rival tops 10% — with monthly visits hitting a record 61.35 billion in September and weekly active users at 1.2 billion. What it means: The “ChatGPT is losing share” story is dead for now — scale is compounding, and challengers need a new angle, not a better chatbot.</description><pubDate>Mon, 09 Nov 2026 13:32:00 GMT</pubDate></item>
<item><title>Google ships four TPU variants in one year, ramping up Nvidia rivalry</title><link>https://the-decoder.com/google-ships-four-tpu-variants-in-one-year-as-it-ramps-up-competition-with-nvidia/</link><guid isPermaLink="true">https://the-decoder.com/google-ships-four-tpu-variants-in-one-year-as-it-ramps-up-competition-with-nvidia/</guid><description>Google shipped four TPU variants in one year, adding the TPU 8tx — which keeps data encrypted during processing — and the TPU 8i Max, doubling compute and memory on one package, with nearly 100 external customers now on board. What it means: Google's chip strategy is the quiet threat to Nvidia: a credible in-house alternative with real customers.</description><pubDate>Mon, 09 Nov 2026 12:59:00 GMT</pubDate></item>
<item><title>ChatGPT bans accounts en masse with no warning — including Pro users</title><link>https://www.leiphone.com/category/zaobao/Z2uc8AJRp0PQO7Kb.html</link><guid isPermaLink="true">https://www.leiphone.com/category/zaobao/Z2uc8AJRp0PQO7Kb.html</guid><description>OpenAI banned large numbers of ChatGPT accounts without warning on Nov 8, including $200-a-month Pro subscribers, who lost access to years of chat history with no prior notice. What it means: A reminder that your AI workspace is rented, not owned — export anything you can't afford to lose.</description><pubDate>Mon, 09 Nov 2026 10:32:00 GMT</pubDate></item>
<item><title>Baidu: self-improving ERNIE Evolve already builds its newest models</title><link>https://www.leiphone.com/category/industrynews/FO9Dx6HKj5Bvno5y.html</link><guid isPermaLink="true">https://www.leiphone.com/category/industrynews/FO9Dx6HKj5Bvno5y.html</guid><description>Baidu disclosed that its recursive self-improvement system, ERNIE Evolve, is already used in developing its newest models, and said Wenxin 5.5 Flash will launch soon. What it means: Self-improving models are moving from research papers to production roadmaps — watch whether rivals disclose the same.</description><pubDate>Mon, 09 Nov 2026 09:48:00 GMT</pubDate></item>
<item><title>ByteDance ships Seedream 5.5 at half the price of its predecessor</title><link>https://www.leiphone.com/category/industrynews/CoJod8Z9X1pDj1sv.html</link><guid isPermaLink="true">https://www.leiphone.com/category/industrynews/CoJod8Z9X1pDj1sv.html</guid><description>ByteDance released Seedream 5.5, its image model with faster generation, more stable multi-round editing and transparent-asset support, at about ¥0.28 per 2K edit — under half the cost of Seedream 5.0 Pro. What it means: Image generation is now a cost war: ByteDance is halving prices while improving quality.</description><pubDate>Mon, 09 Nov 2026 09:41:00 GMT</pubDate></item>
<item><title>Tencent's Hunyuan Hy4 joins the strongest open-weight models</title><link>https://www.qbitai.com/2026/11/507843.html</link><guid isPermaLink="true">https://www.qbitai.com/2026/11/507843.html</guid><description>Tencent's Hunyuan Hy4 is now among the strongest open-weight models in independent testing, paired with WorkBuddy for multimodal agent tasks. What it means: China's open-weight bench is deep — Hy4 joins DeepSeek, GLM and Qwen in the top tier.</description><pubDate>Mon, 09 Nov 2026 08:44:00 GMT</pubDate></item>
<item><title>Qwen Office 2.0: the first “Team Agent” for multi-person work</title><link>https://www.leiphone.com/category/industrynews/ljjfjnedPmwgYh6Q.html</link><guid isPermaLink="true">https://www.leiphone.com/category/industrynews/ljjfjnedPmwgYh6Q.html</guid><description>Alibaba released Qwen Office 2.0, the first office suite built around “Team Agents” — AI agents that join multi-person workflows instead of working solo. What it means: The office AI race is moving from “one person + one agent” to “whole team + agents” — collaboration, not capability, is the new moat.</description><pubDate>Mon, 09 Nov 2026 07:46:00 GMT</pubDate></item>
<item><title>基元律动发布NeoHorse-1.5-Flash Preview，Agent模型RSI进入第二轮迭代</title><link>https://www.qbitai.com/2026/11/508603.html</link><guid isPermaLink="true">https://www.qbitai.com/2026/11/508603.html</guid><description>11月5日，基元律动（TokenRhythm）发布Agent原生模型NeoHorse-1.5-Flash Preview。 What it means: Fresh from 量子位 — read the original for full context.</description><pubDate>Mon, 09 Nov 2026 05:09:23 GMT</pubDate></item>
<item><title>Nvidia board launches investigation into CEO Jensen Huang</title><link>https://www.theinformation.com/briefings/nvidia-board-launching-investigation-reporting-ceo-targeted-chinese-informant</link><guid isPermaLink="true">https://www.theinformation.com/briefings/nvidia-board-launching-investigation-reporting-ceo-targeted-chinese-informant</guid><description>Nvidia's board opened an independent investigation after reports that CEO Jensen Huang was the target of a foreign intelligence effort. Huang admitted in a company memo that he had an affair. Yale's Jeffrey Sonnenfeld puts the odds of Huang keeping his job at “about 50-50”. What it means: The governance question now matters more than the personal one: investors are watching whether the board finds any risk to Nvidia's secrets — or to its CEO's seat.</description><pubDate>Sun, 08 Nov 2026 20:26:00 GMT</pubDate></item>
<item><title>AI's $300 billion stock binge is testing investors</title><link>https://www.theinformation.com/articles/ai-stock-can-market-absorb</link><guid isPermaLink="true">https://www.theinformation.com/articles/ai-stock-can-market-absorb</guid><description>Tech giants, chip companies and AI cloud providers have sold or priced more than $300 billion of new stock and mandatory convertible securities this year, with roughly another $100 billion in potential sales still on tap. What it means: The AI boom is being financed in public markets at historic scale — the question is whether investors keep absorbing the supply.</description><pubDate>Sun, 08 Nov 2026 16:00:00 GMT</pubDate></item>
<item><title>AI coding agents still need human cleanup on nearly 3 of 4 real tasks</title><link>https://the-decoder.com/ai-coding-agents-still-need-human-cleanup-on-nearly-three-out-of-four-tasks-in-a-real-world-test/</link><guid isPermaLink="true">https://the-decoder.com/ai-coding-agents-still-need-human-cleanup-on-nearly-three-out-of-four-tasks-in-a-real-world-test/</guid><description>On Alibaba's Co-Design Bench — real user requests that already tripped agents up in practice — the best model, GPT-6 Astra, fully solves only 26.4% of tasks, versus 99%+ on older benchmarks. What it means: The benchmark gap is the story: agents look near-perfect in demos and need human cleanup on nearly three of four real tasks. Budget review time, not just tokens.</description><pubDate>Sun, 08 Nov 2026 09:50:00 GMT</pubDate></item>
<item><title>Anthropic pushes ahead with pre-Thanksgiving IPO despite tech selloff</title><link>https://the-decoder.com/anthropic-pushes-ahead-with-pre-thanksgiving-ipo-despite-tech-selloff-accepts-lower-valuation/</link><guid isPermaLink="true">https://the-decoder.com/anthropic-pushes-ahead-with-pre-thanksgiving-ipo-despite-tech-selloff-accepts-lower-valuation/</guid><description>Anthropic will start marketing its IPO to investors next week, aiming to raise $80–100 billion before Thanksgiving at a $1.25–1.75 trillion valuation — well below the $2 trillion it once hoped for — after two weeks of volatile tech trading. What it means: The first pure-play AI lab IPO will price the whole AI boom: a weak debut cools the sector's funding frenzy; a strong one reopens the tap.</description><pubDate>Sat, 07 Nov 2026 17:37:00 GMT</pubDate></item>
<item><title>AI auditors catch most agent misbehavior — but flood reviewers with false alarms</title><link>https://the-decoder.com/ai-auditors-catch-most-agent-misbehavior-but-flood-reviewers-with-false-alarms-new-study-finds/</link><guid isPermaLink="true">https://the-decoder.com/ai-auditors-catch-most-agent-misbehavior-but-flood-reviewers-with-false-alarms-new-study-finds/</guid><description>AI models can spot most misbehavior by other AI agents in logs — GPT-5 caught about 98% of reward-hacking cases in a study by Anthropic researcher Andy Hall — but tuning the search harder floods reviewers with false alarms. What it means: AI oversight of AI works in the lab but not yet in production — false alarms are the bottleneck, not detection.</description><pubDate>Sat, 07 Nov 2026 16:56:00 GMT</pubDate></item>
<item><title>From the front row to the hot seat — what the midterms mean for Big Tech</title><link>https://techcrunch.com/2026/11/06/from-the-front-row-to-the-hot-seat-what-the-midterms-mean-for-big-tech/</link><guid isPermaLink="true">https://techcrunch.com/2026/11/06/from-the-front-row-to-the-hot-seat-what-the-midterms-mean-for-big-tech/</guid><description>After spending two years getting cozy with the White House, Big Tech is suddenly trying to make friends with Democrats, and the open question is whether that image makeover can hold up against what states, local communities, and the markets are about to do. What it means: Fresh from TechCrunch AI — read the original for full context.</description><pubDate>Sat, 07 Nov 2026 04:16:43 GMT</pubDate></item>
<item><title>OpenAI raises another $11 billion at a $1.25 trillion valuation</title><link>https://techcrunch.com/2026/11/06/openai-reportedly-raises-an-additional-11-billion-at-a-1-25-trillion-valuation/</link><guid isPermaLink="true">https://techcrunch.com/2026/11/06/openai-reportedly-raises-an-additional-11-billion-at-a-1-25-trillion-valuation/</guid><description>OpenAI has closed the first part of its latest round, raising $11 billion at a $1.25 trillion valuation, and will brief investors next week on third-quarter results. What it means: Two $1.25T valuations in one week — Anthropic and OpenAI are now priced identically, and the market is about to decide whether both can be right.</description><pubDate>Fri, 06 Nov 2026 22:53:00 GMT</pubDate></item>
<item><title>Thinking Machines' new model takes the lead in western open-weight AI</title><link>https://www.zdnet.com/innovation/thinking-machines-inkling-2-small-open-weight/</link><guid isPermaLink="true">https://www.zdnet.com/innovation/thinking-machines-inkling-2-small-open-weight/</guid><description>Thinking Machines' new open-weight model now leads the Western open-weight field, giving developers outside China a competitive open alternative. What it means: Open-weight competition is no longer a China-only story — Western labs are now fighting for the same developers.</description><pubDate>Fri, 06 Nov 2026 17:08:00 GMT</pubDate></item>
<item><title>Ramp data shows AI spending stalling just as Anthropic nears its IPO</title><link>https://techcrunch.com/2026/11/06/ramp-data-shows-ai-spending-stalling-just-as-anthropic-nears-its-ipo/</link><guid isPermaLink="true">https://techcrunch.com/2026/11/06/ramp-data-shows-ai-spending-stalling-just-as-anthropic-nears-its-ipo/</guid><description>AI usage keeps climbing, but spending has stalled, according to Ramp data — price cuts and smarter cost controls among the biggest spenders could challenge investor assumptions about AI hypergrowth. What it means: Usage up, spend flat: the market is learning to do more AI per dollar, and that changes the revenue math for every AI vendor.</description><pubDate>Fri, 06 Nov 2026 15:54:00 GMT</pubDate></item>
<item><title>I watched a humanoid robot clean my apartment</title><link>https://www.technologyreview.com/2026/11/06/1148090/i-watched-a-humanoid-robot-clean-my-apartment-it-felt-like-a-taste-of-the-future/</link><guid isPermaLink="true">https://www.technologyreview.com/2026/11/06/1148090/i-watched-a-humanoid-robot-clean-my-apartment-it-felt-like-a-taste-of-the-future/</guid><description>A humanoid-robot cleaning service in San Francisco let a MIT Technology Review writer watch a robot clean her apartment — an early taste of domestic robots doing real chores. What it means: Humanoids are leaving the demo video and entering homes — the cleaning test is the first job that has to work unsupervised.</description><pubDate>Fri, 06 Nov 2026 10:00:00 GMT</pubDate></item>
<item><title>Ads in chatbots have arrived — and Anthropic is the holdout</title><link>https://www.wired.com/story/ads-in-chatbots-have-arrived-heres-whats-working-and-whats-not</link><guid isPermaLink="true">https://www.wired.com/story/ads-in-chatbots-have-arrived-heres-whats-working-and-whats-not</guid><description>Ads have arrived inside AI chatbots, with most of the industry racing to win over advertisers — while Anthropic holds its vow to keep ads out of Claude. What it means: The free-chatbot business model is being decided now: ads or subscriptions — and the choice will shape what these products become.</description><pubDate>Thu, 05 Nov 2026 21:37:00 GMT</pubDate></item>
<item><title>Google ships Gemini Omni 1.2 Flash: faster, cheaper, longer clips</title><link>https://deepmind.google/blog/gemini-omni-12-flash-fast-and-cost-effective-with-longer-clip</link><guid isPermaLink="true">https://deepmind.google/blog/gemini-omni-12-flash-fast-and-cost-effective-with-longer-clip</guid><description>Google released Gemini Omni 1.2 Flash, a faster and cheaper multimodal model that handles longer video clips. What it means: The “fast and cheap” tier is where most real usage lives — Flash-class models decide who wins the API volume war.</description><pubDate>Thu, 05 Nov 2026 17:05:00 GMT</pubDate></item>
<item><title>OpenAI launches GPT-6 Cyber, a hacking AI that stays in bounds</title><link>https://openai.com/index/introducing-gpt-6-cyber</link><guid isPermaLink="true">https://openai.com/index/introducing-gpt-6-cyber</guid><description>OpenAI launched GPT-6 Cyber, a model built to find exploitable vulnerabilities and develop fixes while staying inside authorized scope. In independent tests, DeepSeek and GLM's hacking skills were “not far behind” OpenAI's most advanced cyber model. What it means: Offense and defense just got the same tooling — security teams should assume attackers already have this capability.</description><pubDate>Thu, 05 Nov 2026 09:00:00 GMT</pubDate></item>
</channel></rss>