<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0"><channel><title>HotAI.news · AI News</title><link>https://hotai.news/en/news/</link><description>AI news and short guides</description><item><title>Vals AI research: The political stance of Chinese AI models is more complex than expected and not merely that of government mouthpieces | AI News | HotAI.news</title><link>https://hotai.news/en/news/items/3bb54722bda30d70/</link><description>Vals AI research: The political stance of Chinese AI models is more complex than expected and not merely that of government mouthpieces</description></item><item><title>Terence Tao criticizes AI methods for solving math problems: The issue lies not in proof correctness.</title><link>https://hotai.news/en/news/items/dfd453a7e04c9f5b/</link><description>Rohan Paul shares Terence Tao's video perspective: Mathematicians' dissatisfaction with AI solving math problems is unrelated to the correctness of the proofs. Tao notes that while activity surrounding the Navier-Stokes equations is currently chaotic and will eventually recover to understanding the proof, this approach is highly undesirable. He calls this pattern irresponsible automated open-pit mining of problems—achieving immediate goals at the expense of the field's future development. The full video is on the New Scientist YouTube channel.</description></item><item><title>OpenAI imposes payment card and device limits on ChatGPT paid subscriptions</title><link>https://hotai.news/en/news/items/69c65a918d331d2b/</link><description>OpenAI has added payment card and device limits to ChatGPT paid subscriptions. Once a single card or device is bound to too many subscriptions, new ones cannot be created; when the limit is reached, the message 'Device limit reached' appears.</description></item><item><title>Google launches StreamVibe video agent to let AI decide when to speak up and help.</title><link>https://hotai.news/en/news/items/6c0f302e650f09d6/</link><description>Google, together with teams from Stanford and Oxford, launched the StreamVibe video agent. It can distinguish between what users say explicitly, what they haven't said yet, and changes in tasks mid-way, autonomously deciding when to speak up. Its StreamVibe-VL model is based on Qwen3.5-9B, with response timing and quality improved by 10.7% compared to other models, and its memory system improving active interaction by another 12.3%. Tests show that after videos exceed five minutes, all models struggle to grasp the right timing, with key content metrics based on their own historical answers dropping as much as 61.5%.</description></item><item><title>CADI Research Institute report: China's AI inference data volume surpasses training data for the first time, with average monthly token calls exceeding 175 trillion in June.</title><link>https://hotai.news/en/news/items/50ac0438ccdd71ee/</link><description>A report released yesterday by the CADI Research Institute titled 'Top Ten Trends in Innovation and Development of New Domains and New Quality' shows that China's artificial intelligence inference data volume has exceeded its training data volume for the first time, signaling a shift from model training and R&amp;D to scaled industrial application. As of June 2026, average monthly token calls surpassed 175 trillion, representing an increase of over 1,000 times compared to early 2024. Domestic computing power and large models have been deployed in over 1,000 application scenarios across 20 industries including steel and automotive, with the cumulative construction of more than 500 excellent-level smart factories.</description></item><item><title>The Ministry of Industry and Information Technology's Academy for the Application of New Technologies (CAICT) reports that China's AI inference data volume has surpassed its training data volume for the first time. As of June, the average daily token calls exceeded 175 trillion.</title><link>https://hotai.news/en/news/items/3b5c025e0a7aa2ef/</link><description>According to today's report by CCTV, the China Electronics Industry Institute (CCID) released 'Top Ten Trends in Innovation of New Domains and New Quality Productive Forces' yesterday. The data shows that China's AI inference volume has exceeded its training volume for the first time, marking a shift from the model training and R&amp;D phase to the stage of large-scale industrial application.</description></item><item><title>ChatGPT to merge Chat and Work features, sharing a single token budget starting in December</title><link>https://hotai.news/en/news/items/3e3d91cd906d037c/</link><description>According to OpenAI's updated support page, the Plus, Pro, and Business plans will share a unified token budget between ChatGPT and Codex starting in December. Work's agent features are being integrated into the main interface.</description></item><item><title>David Sacks reports that Anthropic disclosed in August that AI has taken on 26% of the research tasks for next-generation models.</title><link>https://hotai.news/en/news/items/1cfd1caf766d8769/</link><description>Rohan Paul shared David Sacks' statement from the All-In Podcast. In a blog post released by Anthropic in August, it was revealed that 26% of tasks related to building next-generation models have already been completed by RSI (Recursive Self-Improvement). Sacks added that this figure reflects only August data, and progress for October remains unknown.</description></item><item><title>What is AI bringing to math education? Wolf Prize winner promises to end boring teaching methods</title><link>https://hotai.news/en/news/items/e9bd11d289197f8a/</link><description>Could machines do mathematics? This once-philosophical question has sparked global mathematical debate and deep discussion in 2026. 'I hope that as artificial intelligence becomes increasingly adept at handling tedious and dull tasks, we can...'}, {</description></item><item><title>Ali Cloud Qwen AI Glasses OTA Update: Adds Call Translation, Media Translation, and More</title><link>https://hotai.news/en/news/items/c6f9008c6eb9e12d/</link><description>The latest update introduces cross-border calling, overseas podcasts, foreign language online courses, and real-time translation of major films. Noise-canceling voice recognition has been optimized for noisy environments, while simultaneous interpretation is now more accurate and timely. Translations can be displayed in real time on the Qwen AI Wearable App.</description></item><item><title>Silicon Flow's Hong Kong IPO filing application status shows as "Received," aiming to become the first stock in the "AI Token Factory" sector.</title><link>https://hotai.news/en/news/items/3dfacdff3a56c803/</link><description>On October 30, information disclosed on the China Securities Regulatory Commission website showed that Beijing Silicon Flow Technology Co., Ltd. (short for "Silicon Flow") had its Hong Kong stock issuance and listing filing status set to "Received," with the material receipt date being October 1, 2026.</description></item><item><title>Nace AI has open-sourced its 27B decision-making model Drex Max v1, which can run locally on Apple Silicon.</title><link>https://hotai.news/en/news/items/46c34ec0936652c0/</link><description>Nace AI has open-sourced its decision-making model Drex Max v1 under the Nace.AI Open RAIL-M license. The weights include bf16, Q8_0 GGUF, and affine Q4 checkpoints, with support for llama.cpp and an Ollama fork.</description></item><item><title>DeepSeek open-sources the coding agent dsh-superagent via Router-Worker</title><link>https://hotai.news/en/news/items/e92eb9392c42b99a/</link><description>The DeepSeek Harness team has open-sourced the coding agent prototype dsh-superagent v0.1.0 under an MIT license. Designed for a single GitHub repository, it can monitor requests from both GitHub and Feishu to generate parallel tickets.</description></item><item><title>Greenhouse survey: Gen Z job seekers worry that AI-polished resumes will fail AI screening</title><link>https://hotai.news/en/news/items/c7f9448321d8f922/</link><description>A Greenhouse survey of over 1,100 job seekers in the UK and Ireland found that 81% were rejected after submitting only their resumes without any interaction with recruiters. Additionally, 42% of candidates had taken AI interviews in the past 12 months, with 57% believing such interviews offer a chance to demonstrate their skills.</description></item><item><title>Exin Therapeutics enables AI agents to perform brain surgery for epilepsy treatment in mice.</title><link>https://hotai.news/en/news/items/055c9f04e98741cd/</link><description>Biotech startup Exin Therapeutics claims it has enabled an AI agent to perform brain surgery for epilepsy treatment in mice. CEO Gabriel Ocana-Santero shared the process on X.</description></item><item><title>Amidst the flood of AI-generated resumes, Gen Z job seekers are growing concerned that their applications may look too artificial to pass initial screening.</title><link>https://hotai.news/en/news/items/e5f4875a0b14a721/</link><description>Research by recruitment platform Greenhouse reveals that most job seekers receive rejection notices from AI screening systems. AI-polished resumes show high similarity levels, and both recruiters and applicants agree the current process has issues and urgently needs a fairer competition mechanism.</description></item><item><title>Rumors claim 'wild monkeys in Xinjiang's Heimu scenic area descend to snatch mobile phones,' but the official response states: 'This is AI-generated; there are no monkey habitat activities in the area.'</title><link>https://hotai.news/en/news/items/81c713a5d1c4c748/</link><description>After verification by the Bu'lin County Cyberspace Administration of Xinjiang in coordination with local forestry and grassland authorities, it has been confirmed that rumors claiming wild monkeys descended from Heimu to snatch mobile phones are AI-generated falsehoods. There is no record of wild monkey habitats in the area, and the false content has been ordered removed.</description></item><item><title>AI performs brain surgery itself; startup succeeds in using agent to treat epilepsy in mouse</title><link>https://hotai.news/en/news/items/60d96facd3f17142/</link><description>Exin Therapeutics, a biotech startup focused on next-generation neurotherapies, claims this week that its AI agents have successfully performed brain surgery on a mouse.</description></item><item><title>Grok will add invisible watermarks to generated text starting November 7 to comply with the EU AI Act.</title><link>https://hotai.news/en/news/items/909bf69d0077aaa2/</link><description>SpaceXAI announced that it will add invisible watermarks to Grok-generated text starting November 7 in response to requirements under the EU's Artificial Intelligence Act.</description></item><item><title>In response to the EU AI Act, text generated by the Grok model will include invisible watermarks</title><link>https://hotai.news/en/news/items/f3a550b2ed7d8ca1/</link><description>xAI has added invisible watermarks to text generated by the Grok model in response to EU AI Act requirements. Watermarking is now active globally across all APIs. Administrators can disable it for specific models individually. The watermark is integrated into the token selection process.</description></item><item><title>The US Department of Justice launched Operation Ghostbusters, an investigation into AI-generated fake students defrauding over $200 million in student loans.</title><link>https://hotai.news/en/news/items/59917e9503107aeb/</link><description>The US Department of Justice announced the results of Operation Ghostbusters on October 31 in local time. The operation investigated fraud against the Federal Student Aid (FAFSA) program. Fraudsters used stolen identities to apply for student loans and employed AI to complete applications, making it appear that fake students were enrolled in online courses. Some remaining funds flowed into the fraudsters' accounts and were transferred overseas, leaving multiple victims who never attended college with debt.</description></item><item><title>Roblox's AI game count surpasses 70,000; over 80% of the top 100 September releases used AI assistance</title><link>https://hotai.news/en/news/items/193ebddd10c2a316/</link><description>Roblox's Q3 2026 financial report reveals that over 70,000 AI-generated games built on its platform have been released. More than 80% of the top 100 most-played games in September utilized AI for development.</description></item><item><title>A small guide to managing Grok Bots: keep the team lean, focus on skills, and leverage voice features</title><link>https://hotai.news/en/news/items/d9b3c10f171b2004/</link><description>A guide for managing the Grok Bot team suggests that fewer, well-performing bots are easier to manage than many. It recommends letting bots save their methods as skills to maintain long-term consistency. The guide also advises connecting multiple tools and using Vault so bots can have their own email addresses and accounts. It suggests identifying manual or copy-paste tasks for the bots to handle. The author recommends having users verbally dump their ideas, then letting the bot rephrase them in its own words to confirm understanding.</description></item><item><title>The US Department of Justice is urgently investigating a case where AI-generated fake students were used to fleece $200 million in student loans.</title><link>https://hotai.news/en/news/items/1ac412df197f30d5/</link><description>According to that agency, bad actors use stolen identities to apply for student loans and leverage AI to complete applications, making it appear as if fake students are enrolled in online courses.</description></item><item><title>Roblox platform AI game count exceeds 70,000; over 80% of the top 100 titles use AI-assisted development</title><link>https://hotai.news/en/news/items/328708a2a1038352/</link><description>According to Roblox's Q3 FY2026 report, the Build tool has over 500,000 users. The platform hosts more than 70,000 AI games, with over 80% of the top 100 games using AI-assisted development. Two games created almost entirely by AI have entered the top 20 on the concurrent players list.</description></item><item><title>OpenAI to unify ChatGPT and Codex under a single token quota starting this December</title><link>https://hotai.news/en/news/items/90fa131b3c6b313b/</link><description>Citing an early update to OpenAI's help documentation, the author notes that starting in December, ChatGPT usage and Codex consumption for Plus and Pro users will be merged into a single token quota; Instant chat remains unaffected. The author reviews the history of per-request billing, the operational fluctuations where the Codex Pro monthly quota first decreased then recovered, and community reactions to GitHub issue protests, concluding that the merger is related to computing cost pressures under IPO expectations.</description></item><item><title>OpenRouter tutorial: Build an AI code review script using the Responses API that can verify bugs in a Linux sandbox.</title><link>https://hotai.news/en/news/items/fadab16d52397956/</link><description>OpenRouter tutorial: Build an AI code review script using the Responses API that can verify bugs in a Linux sandbox, allowing the model to run checks, confirm issues, and return V4A patches.</description></item><item><title>OpenAI announces that starting in December, Plus, Pro, and Business users will share quotas between ChatGPT and Codex</title><link>https://hotai.news/en/news/items/69527890b1d3a635/</link><description>OpenAI announced that starting in December, as Work's agent capabilities are integrated into the main ChatGPT experience, Plus, Pro, and Business subscribers will share a unified usage quota between ChatGPT and Codex.</description></item><item><title>Sources indicate Spotify plans to launch an AI music remix service in early 2027; subscribers may need to pay an additional $4 to $5 monthly</title><link>https://hotai.news/en/news/items/a40df12031c686ac/</link><description>According to The Financial Times, Spotify plans to launch an AI music remix service in the first quarter of 2027, potentially debuting around February near the Grammy Awards. Subscribers may need to pay an additional $4 to $5 per month.</description></item><item><title>Parameter Labs reveals the attack chain for taking over a ChatGPT account</title><link>https://hotai.news/en/news/items/df8ebd27f1023141/</link><description>Simon Willison reports on a ChatGPT sandbox escape vulnerability chain disclosed by Parameter Labs. Untrusted code within an HTML preview iframe can trigger MathLive's virtual keyboard via postMessage, injecting innerHTML content. This is then reflected and executed as JavaScript through a YouTube oEmbed endpoint permitted by Content Security Policy (CSP), potentially allowing attackers to steal private data from logged-in users or perform actions on their behalf.</description></item><item><title>OpenAI: Starting in December, ChatGPT will share quotas with Codex</title><link>https://hotai.news/en/news/items/f5615d738f697922/</link><description>Usage on ChatGPT and Codex will now be uniformly measured in tokens. Actual consumption rates depend on the selected model and task complexity, while Instant Chat does not consume any quota.</description></item><item><title>Reports indicate Spotify plans to launch an AI music remix service early next year, allowing users to create their own song remixes and covers.</title><link>https://hotai.news/en/news/items/9604254466c05313/</link><description>Spotify is actively advancing copyright licensing negotiations and working to convince musicians to embrace new technologies.</description></item><item><title>Surging AI demand: Intel rapidly retools multiple CPU fabrication plants, shifting from desktop processors to data center chips</title><link>https://hotai.news/en/news/items/ffd97caed0f4e1c5/</link><description>Intel is reallocating some production capacity from client CPUs to data center CPUs, expanding Fab 24 in Ireland to produce more Xeon chips, beginning equipment installation at Fab 62 in Arizona ahead of schedule, and committing to the production of 18A chips. The company is also expanding advanced packaging capabilities in New Mexico and Malaysia.</description></item><item><title>David Sacks, Chair of the White House Science and Technology Policy Council, suggests Anthropic may become the greatest company in capitalist history</title><link>https://hotai.news/en/news/items/c5502762b1add360/</link><description>David Sacks, Chair of the White House Science and Technology Policy Council, analyzed Anthropic's S-1 filing on the All-In podcast, suggesting it could become the greatest company in capitalist history and potentially the highest-return venture capital investment.</description></item><item><title>Chairman of the US White House Science and Technology Advisory Committee: Anthropic could become the greatest company in capitalist history.</title><link>https://hotai.news/en/news/items/e07d57941fe17931/</link><description>Sax analyzed Anthropic's S-1 filing on the All-In podcast, noting that its IPO financing needs were lower than expected, with ample cash and remarkable capital efficiency. #Anthropic #AIInvestment Seven founders control 50.1% of voting rights through Class F shares; the company expects annual recurring revenue (ARR) to reach $100 billion in 2026.</description></item><item><title>OpenAI official documentation announces that starting in December, Plus, Pro, and Business plan users will share token usage between ChatGPT and Codex.</title><link>https://hotai.news/en/news/items/ca39afd09e04bde4/</link><description>OpenAI official documentation states that starting in December, ChatGPT and Codex on Plus, Pro, and Business plans will share the same token usage quota. This is because OpenAI integrated the agent features built on Codex in July into the main ChatGPT experience.</description></item><item><title>A paper from the University of Illinois introduces ClaimNov to enhance novelty checks in AI research workflows.</title><link>https://hotai.news/en/news/items/2fd52c9adc870174/</link><description>The team's paper proposes ClaimNov, which splits proposals into individual claims and compares them against existing literature before review. This maintains a plagiarism detection rate of 97% while reducing false negatives to 1.5%. The authors suggest teams filter ideas based on claim-level scoring and retain proposals that build incrementally on known results.</description></item><item><title>Anthropic has added watermarks to four older Claude models; two of these can be bypassed using sampling parameters.</title><link>https://hotai.news/en/news/items/37faf7453f89fc12/</link><description>Anthropic announced it will extend text watermarking to Claude Mythos 5, Claude Opus 4.7, Claude Opus 4.6, and Claude Sonnet 4.6 to meet the December deadline of the EU AI Act.</description></item><item><title>Grok Bot engineering head shares methods for building high-quality bots and managing their teams</title><link>https://hotai.news/en/news/items/d217930b9d27da8f/</link><description>Lauren Tan, head of the Grok Bot engineering team, shares a guide on building high-quality bots and managing their teams, along with an entry point at https://x.ai/bot/_jOdbfkB16zxu7MRcmReE.</description></item><item><title>xAI engineers share Grok Bot tips: routines and automation</title><link>https://hotai.news/en/news/items/879ca606aadeed7d/</link><description>xAI's Lauren Tan shares advanced Grok Bot usage: routines make bots more proactive, supporting scheduled tasks (such as checking the issue tracker every morning) and subscription-based triggers (such as automatically summarizing and writing to Notion when receiving a specific customer email).</description></item></channel></rss>
