NAll news

All news

Friday66 items

  1. CTechOpinion
    AI score70

    Anthropic's $2 trillion IPO question: how much growth can investors buy

    As Anthropic prepares for its IPO, leaked data shows its third-quarter revenue was $16.8 billion, up 45% quarter-over-quarter, exceeding the $16.3 billion total for the first half of 2026; the first half of 2026 had a loss of $80.7 billion, of which $72 billion was an accounting loss from debt-to-equity conversion, and even after excluding that, the loss was still $8.7 billion. The article points out that its revenue is highly concentrated among a few customers; in the first half of the year, two customers contributed 20% of revenue, and it has signed AI infrastructure contracts totaling $600 billion over the next decade, with another $150 billion under negotiation. The article uses SpaceX's $1.77 trillion valuation at IPO and OpenAI's rumored $1.4 trillion private valuation as references to discuss whether a $2 trillion pricing can be accepted by the market.

    #Industry news#OpenAI

  2. HuxiuOpinion
    AI score58

    AI capex panic pushes down US Treasury yields, market lowers expectations of Fed rate hikes

    The AI capex panic triggered a plunge in US Treasury yields, with the 10-year Treasury yield falling 19 basis points in a single day, the largest one-day drop since the Silicon Valley Bank crisis in early 2023, and the market now sees a 24% probability that the Fed will not raise rates this year. After Meta gave guidance that capex would require "hundreds of billions of dollars," its stock fell nearly 7%, CoreWeave dropped 10%, and credit spreads across the computing infrastructure sector widened across the board. The article argues that the Fed faces a dilemma between fighting inflation and avoiding pricking the AI bubble, while international oil prices hitting their highest level since 2008 provide ammunition for the hawks.

    #Industry news#English text is being prepared.#English text is being prepared.

  3. Latent SpaceOpinion
    AI score73

    Latent Space summary: Black October security incident wave

    Latent Space compiles the vulnerabilities and intrusion incidents that emerged in a concentrated burst over roughly 48 hours from October 21 to 22 into Black October, covering the PHP2Shell unauthenticated RCE, a CVSS 10 vulnerability present in GitLab versions since 2014 (fixed in 19.5.1), LiteLLM 1.107.0 carrying a password stealer, and the eMudhra CA having weak keys and exiting the CA business. The article also mentions that Microsoft attributed the ConnectWise supply chain attack to Storm-0940, which used the open-source agent penetration testing framework ARTEX to launch roughly 2,900 login attempts per minute against more than 2,250 Microsoft 365 tenants. The author also includes the opposing view, which holds that this looks more like a defensive audit and that there is currently no evidence of large-scale exploitation.

    #Industry news#Safety / alignment#Agents

  4. Business Insider TechOpinion
    AI score70

    AI discovers two critical vulnerabilities in PHP, industry warns cybersecurity faces a vulnerability wave

    Business Insider summarizes statements from multiple executives and analysts on AI exacerbating cybersecurity risks: AI helped discover two critical vulnerabilities in PHP infrastructure, and PHP is used by as many as 70% of websites. Anthropic Deputy Chief Information Security Officer Jason Clinton says a vulnerability wave is imminent, says the situation will worsen first over the next 4 days, and calls on enterprises to shore up basic security first; Salesforce CEO Marc Benioff responds that this is a red alert. Investor Deedy Das says non-technical people can also use AI models to locate common vulnerabilities; Box CEO Aaron Levie believes agents will end security through obscurity and predicts rising demand for cybersecurity roles.

    #Industry news#Safety / alignment#Agents

  5. WallstreetcnOpinion
    AI score66

    Musk plans to integrate SpaceX and Tesla to build a full-stack AI empire

    Wall Street See News compiled SpaceX's quarterly data and expansion plans, stating that its third-quarter revenue was $14.6 billion, up 161% year-over-year, and that it recorded positive operating profit for the first time. The article says SpaceX brings about 220,000 NVIDIA GB300 chips online per month and plans to build its own gas turbines for power supply; Musk has confirmed that special committees of independent directors at SpaceX and Tesla are discussing a merger. The article also mentions that SpaceX's third-quarter free cash flow was about negative $10 billion, with capital expenditure reaching $19.6 billion.

    Same story · 13 reports

    #Industry news#English text is being prepared.#English text is being prepared.

  6. Shanghai ObserverOpinion
    AI score62

    AI product images flood e-commerce: merchants save ninety percent of costs, buyers struggle to tell real goods

    Shanghai Observer reports that during Double Eleven, more and more e-commerce merchants use AI to generate product main images, model videos and buyer-show comments in review sections, making it hard for consumers to see the real material and fit. Some merchants say the cost of a set of product images has dropped from one to two thousand yuan to ten to twenty yuan, a reduction of over 90%, and the image production cycle has shortened from two weeks to the same day; meanwhile, some buyers use AI to fabricate defect images to apply for refunds only. The report mentions that the Measures for Labeling AI-Generated Synthetic Content took effect on September 1, 2025, and Taobao launched a special governance campaign against fake AI images in March 2025, intercepting nearly 100,000 fake images within a month, but merchants believe the platform's recognition capability is still limited.

    #English text is being prepared.#Image generation#Industry news

  7. Latent SpaceOpinion
    AI score62

    Claude Mods: Flipping the Agent Framework into a Human-Oriented Attention Interface

    Latent Space author Dan McAteer introduces Claude Mods, a plugin that uses TypeScript or JavaScript functions to respond to prompts, tool calls, and interface rendering events in Claude Code, and can change execution and rendering behavior in the CLI and desktop. The article compares OpenAI's ChatGPT Intelligent UI with Cursor's PR Routing & Approval, arguing that the core difference among the three lies in who designs the attention interface. The author also lists issues encountered in actual use: Next Steps Supervisor triggered in 99 of 111 rounds, each check consumes about 5–6k input-equivalent tokens, and mods do not run in a sandbox.

    #English text is being prepared.#Agents#English text is being prepared.

  8. Latent SpaceOpinion
    AI score51

    Modal CEO Erik Bernhardsson on why AI compute is not rectangular

    In an interview, Modal CEO Erik Bernhardsson explained that traditional compute contracts are signed for a fixed number of GPUs and a fixed term, whereas real workloads grow, spike, and fall back over time; Modal's approach is to aggregate demand from thousands of customers into a smoother pool and then procure capacity centrally. He disclosed that after a $500 million Series C, Modal's annualized revenue exceeded $1 billion, it completed a $750 million Series D at a $16 billion valuation, and it has launched nearly 3 billion sandboxes. The sandbox business was once considered for shutdown, but due to growing demand from coding agents, background agents, and reinforcement learning, it now accounts for about one-third of its revenue.

    #Industry news#Agents#English text is being prepared.

  9. Business Insider TechOpinion
    AI score55

    Sundar Pichai says Google is still far from the endgame of self-improving AI

    Google CEO Sundar Pichai said on the Big Technology Podcast that Google is still far from the endgame of recursive self-improvement (RSI); at present, AI agents are only assisting researchers in various stages of model development, and he expects this process to continue advancing next year. Previously, Google co-founder Sergey Brin said that the Flash series models and Gemini 4 show that AI is helping to build itself, and that Google is beginning to get on the RSI track. Pichai also mentioned that code inside Google is already deeply affected by AI, with code rollbacks reduced by one third.

    #English text is being prepared.#Google#English text is being prepared.

  10. Phoenix TechnologyOpinion
    AI score56

    DeepSeek-powered AI Agents take over Bilibili comment sections

    The open-source plugin dsh-bili-scroll uses DeepSeek Harness to let DeepSeek automatically find Bilibili videos, read subtitles, and post comments; accounts uniformly set to the "Big Fat Fish" persona have appeared in large numbers in the tech section's comment areas. The plugin disables auto-commenting by default; once enabled, it posts one comment per round by default, can be set to a maximum of 20, and limits commenting to once per video. Some netizens subsequently open-sourced the browser plugin dafeiyu-cleaner, which identifies and locally hides these accounts based on nickname, avatar, and comment characteristics; the developer says it identified about 85% of 284 test accounts.

    #English text is being prepared.#Agents#Open-source ecosystem

  11. Tessl: Product and Engineering BlogFirst-partyOpinion
    AI score58

    When agents write code, who should get deployment permissions?

    The Tessl blog proposes that agents can write code and suggest fixes, but should not have permission to deploy to production; deployment permissions should be given to the CI/CD pipeline. The article cites a case from the Cline repository: a malicious issue title can induce its triage agent to install a package from an attacker's fork, and via a shared build cache steal an npm token to publish an unauthorized version. The author advocates treating everything an agent reads as untrustworthy, and replacing manual code review with short-lived OIDC identities, hash-locked dependencies, and an independent verifier.

    DevCon NYC

    #Safety / alignment#Agents#English text is being prepared.

  12. Shao MengOpinion
    AI score62

    How OpenAI defines FDE: consolidating customer-customized Harness into a unified Codex

    OpenAI's FDE team was established in early 2025, with the goal of filling the deployment gap between model capabilities and customers' real-world constraints; the approach is to first get results into production, then consolidate customized solutions into reusable products. In the early days, every use case required extensive customized integration; after Codex became stronger, the team rebuilt the application on Codex, turning customized applications into plugins and skills based on Codex primitives, and self-reports that about 90% of the customized code for each project was removed. In a semiconductor customer case, about 15 FDEs covered 10+ use cases, self-reporting about 48% time savings in targeted test generation and about 22% median performance improvement in kernel optimization; in a telecom customer service voice case, the key evaluation rose from about 60% to 90%+ in about six months.

    #Industry news#Agents#English text is being prepared.

  13. 36Kr · AIOpinion
    AI score53

    Exiting after just 188 days, big-company AI products are accelerating their reshuffle

    36Kr sorted through the product moves of eight companies—Tencent, Alibaba, Baidu, ByteDance, and Moonshot AI, Zhipu, MiniMax, DeepSeek—across ten AI tracks over four years, finding that competition cycles have shortened, product lifecycles have grown shorter, and renaming, integration, and shutdowns have become the norm. Tencent QClaw went from public beta on March 20 to announcing service shutdown arrangements on September 24, just 188 days, with user data to be migrated to WorkBuddy; Alibaba's Qwen announced the upcoming release of its personal assistant product Qwen Go!. Alibaba and ByteDance cover all ten tracks, Tencent and Baidu each cover nine, and DeepSeek covers only two.

    #English text is being prepared.#Agents#Industry news

  14. 36Kr · AIOpinion
    AI score57

    AI mass-generates e-commerce videos: a 5-person team produces over a thousand per day, at under 5 yuan each

    Yunying Research Society analyzes that AIGC is taking over the mass-production stage of e-commerce recommendation videos; a team of about 5 people can generate over a thousand videos in a few minutes, at under 5 yuan each, while manual shooting costs about 100 to 200 yuan per video. The specific approach is to first break out fixed blocks and variable blocks to form a template, then use low-cost models such as Grok Video and the Seedance entry version to test hooks and directions, and after getting data, switch to Veo, Sora, and the Seedance HD version to produce finished products. The article also mentions that Douyin, Xiaohongshu, and Video Accounts all tightened AIGC content review this year, and that Ocean Engine reported at the end of August that it had shut down more than 50,000 violating accounts, and holds that once materials are no longer scarce, judgment and aesthetics become the new threshold.

    #English text is being prepared.#Video#Industry news

  15. 36Kr · AIOpinion
    AI score63

    AI comic drama production costs drop by 90%, money flows to computing power and platform revenue sharing

    AI comic drama production fees fell from 15,000 yuan per minute in 2024 to 100,000 yuan per production and 1,000 yuan per minute in 2025, with batch productions even as low as 400 yuan per minute, but part of the money saved shifted to computing power spending. Practitioners say the computing power cost per episode rose from 17 yuan early on to 80 to 100 yuan, and the share of labor dropped from 60% to 30-40%. The author breaks the industry accounts into four: content, platform, IP, and computing power, pointing out that the price producers pay for computing power is set by upstream, content sales revenue is calculated by the platform's revenue-sharing coefficient, and they also have to spend money on paid traffic promotion to buy exposure.

    #English text is being prepared.#Video#Industry news

  16. 36Kr · AIOpinion
    AI score62

    Liang Wenfeng needs his own Tibo: DeepSeek product department integration and the Agent path

    DeepSeek has integrated the Harness with the web, App, and open platform teams into a product department, led by Cui Tianyi, and the official website has posted recruitment positions in two directions: AI product manager and Agent Harness. The article uses OpenAI Codex lead Tibo Sottiaux and the Anthropic Claude Code team as examples to explain how the execution capabilities of coding tools can be extended into a general work task entry point. DeepSeek, while open-sourcing Harness and the inference framework DeepInfer, is also facing cost pressure on caching and transmission brought by long tasks.

    Same story · 3 reports

    #Industry news#Agents#English text is being prepared.

  17. TMTPost · AGIOpinion
    AI score51

    "You've Been Replaced by AI": Short Drama Creators in 2026

    Xiaguang She interviewed multiple overseas short drama practitioners, documenting that in 2026 live-action short drama shooting projects decreased by 30% to 50% year-on-year, and actor audition calls dropped from twenty to thirty per week to single digits. The article mentions that the shooting cost of one overseas live-action short drama is about 150,000 USD, while the cost of an AI short drama may be only one-tenth of that, and it outlines the competitive landscape of three forces: Chinese teams, overseas transformation teams, and AI-native entrepreneurs. The article also states that TikTok's short drama revenue-sharing coefficient was lowered, with the share per ten thousand views falling from about 30 USD to a few USD.

    #English text is being prepared.#Video#Industry news

  18. TMTPost · AGIOpinion
    AI score73

    Meta Q3 earnings: advertising props up 97% of revenue, Muse still burning money

    Meta Q3 revenue was $64.6 billion, up 26% year-over-year, of which advertising contributed $63.06 billion, accounting for 97.6% of total revenue. The personal agent Muse, launched on September 8, surpassed 10 million downloads within 40 days of launch, but high capital expenditure turned free cash flow negative at $2.12 billion, and the full-year Capex guidance floor was raised from $130 billion to $135 billion. Muse currently mainly relies on subscription fees, with shopping being the most frequent use case; more than 25% of users have used the shopping feature, with average spend per transaction exceeding $200.

    #Industry news#Agents#OpenAI

  19. TMTPost · AGIOpinion
    AI score78

    Pichai Reviews Gemini 4 Argon: Why 3.5 Pro Was Cut

    Google CEO Pichai revealed in the "Sources" interview that Gemini 4 Argon processed about 25 trillion tokens in 72 hours of launch, 3 times that of Gemini 3.0 in the same period, and is still rate-limited due to compute constraints. He explained for the first time the reason for cutting Gemini 3.5 Pro: the team found a gap in the model on Agentic Coding, first used the Flash version to complete intensive iterations from 3.5 to 3.8 in about 3 months, while Argon pretraining had already been completed at that time, so they injected the accumulated gains into Argon post-training. He also mentioned that Google internally has entered the RSI stage of researchers and models collaborating, and said the next-generation Gemini 5.0 will be much stronger than Argon.

    #Industry news#Google#DeepMind

  20. TMTPost · AGIOpinion
    AI score70

    Terence Tao and Deng Yu in conversation: AI math results are "toxic waste," the problem lies in how they are released

    Terence Tao and this year's newly minted Fields Medalist Deng Yu discussed on the SAIR Foundation podcast the controversy sparked by OpenAI's mass release of AI math results; Tao called such bulk problem-solving output a kind of byproduct, or even "toxic waste." The two proposed that the boundaries between problem-solving and understanding are diverging, argued that AI results must come with explanations and traceable citations, that mathematicians should be involved at project inception, and introduced the Open Math OS project launched by SAIR, which first integrates existing open-source models and tools and hopes to build its own model in a few months.

    #Industry news#Papers/Research#Open-source ecosystem