NAll news

All news

Saturday86 items

  1. East Money
    AI score64

    Figure All-Hands Meeting Reveals Roadmap: Figure 04 Enters Component Manufacturing, Betting on Large-Scale Pretraining and $3.5 Billion in Compute

    Figure has released a one-hour internal all-hands meeting video. Founder and CEO Brett Adcock disclosed that the fourth-generation robot Figure 04 has completed design lock and that its components have entered the delivery and validation cycle, calling it the largest improvement of any generation. The Index global data collection project supporting model iteration has accumulated 37.5 million human manipulation videos, collecting at a rate of about 81 minutes of video per second, with weekly data spending reaching $8.1 million; Figure has also finalized a compute partnership with Nscale at an initial scale of $3.5 billion, planning to deploy up to 100,000 GPUs based on Nvidia's Vera Rubin platform, going live in the second half of 2027 at the Bastrop, Texas site. Adcock said the team's new model, trained with the largest-scale pretraining data and GPU resources to date, has completed testing in the lab internally codenamed Black Box, but he did not provide evaluation details.

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

  2. IT Home · Artificial Intelligence
    AI score65

    OpenAI says ChatGPT can automatically terminate conversations fixated on violence, and promises to regularly publish community safety reports

    OpenAI said in a blog post on October 30 that ChatGPT can automatically terminate the current conversation when the dialogue shows that a user is fixated on violence or has thoughts of seriously harming others, and the same account may be temporarily frozen after being interrupted for this reason multiple times. OpenAI has also tightened restrictions on reviewing details of past mass attack incidents and limited information related to guns and weapons that teenagers can access. OpenAI promised to regularly publish a "Community Safety Report," with the first issue to be released in early 2027, covering safety data for the second half of 2026, and updated every six months thereafter; in the third quarter of 2026, it had already referred more than 200 case leads to law enforcement.

    Same story · 3 reports

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

  3. English text is being prepared.
    AI score75

    Alibaba: Apple Intelligence users in China can independently choose whether to enable Qwen integration

    Apple announced that the first batch of Apple Intelligence features will launch in China, and it is partnering with Alibaba to integrate Qwen into some experiences. Chinese users will be able to use Qwen's text and image understanding and content generation capabilities in Siri, Writing Tools, and Visual Intelligence. The features will open for testing with the iOS developer Beta on November 4 and will be released to Chinese users with the official release of iOS 27.3 in December. Alibaba emphasized that users can independently choose whether to enable the Qwen integration, and that requests and responses will not be used for model training.

    Same story · 10 reports

    #Industry news#OpenAI#Google

  4. IT Home · AI
    AI score51

    Anthropic Renames Claude Max 20x to Ultra, Verified Accounts Get Five-Hour Session Limit Removed

    Anthropic has adjusted the naming of its Claude individual paid plans: the former Max 20x is renamed Ultra, the price remains $200 per month, and the 5-hour session usage limit is removed for verified accounts, though weekly limits still apply. The former Max 5x is simplified to Claude Max at $100 per month. Ultra offers 10 times the weekly usage of the Pro plan and has priority access to new models and features; it is currently available only as a monthly subscription. The company says verification will be automatic based on account and payment history, and users who do not see the current session bar under Usage in Settings have passed verification.

    Same story · 11 reports

    #Product updates#Anthropic#Industry news

  5. English text is being prepared.
    AI score67

    Google's in-orbit TPU completes AI inference for the first time, Gemma 4 generates a haiku

    Google announced that a data-center-class TPU running in orbit aboard its "Project Suncatcher" prototype satellite has for the first time produced an AI response; the pre-installed DeepMind model Gemma 4, following a prompt from ground engineers, composed a haiku about being in space. The satellite carries four TPUs and can generate up to 1 kilowatt of electrical power; limited by heat dissipation and energy load, it can only be powered on for about 20 minutes at a time, and the test focuses on verifying the chip's stability and error-correction capability under microgravity, severe temperature swings and the space radiation environment.

    Same story · 8 reports

    #Industry news#Google#DeepMind

  6. IT Home · Artificial Intelligence
    AI score80

    Anthropic's IPO filing discloses nearly $900 billion in future off-balance-sheet obligations

    Anthropic recently submitted its IPO prospectus, and the fine print in the filing shows that it plans to invest nearly $900 billion in off-balance-sheet obligations in areas such as servers, buildings, and power, far exceeding the current liabilities reported on its balance sheet. Its balance sheet currently includes $117.5 billion in liabilities and $112 billion in cash. Bloomberg reported that peers such as Alphabet and Meta have also accumulated hundreds of billions of dollars in future debt not yet recognized on their balance sheets, sparking Wall Street concerns about excessive AI spending.

    Same story · 127 reports

    #Industry news#Anthropic

  7. English text is being prepared.
    AI score69

    Malicious Claude Code Plugin Appears in Cursor's Official Marketplace, Affecting About 5,000 Developers Over Roughly 11 Hours

    Cursor's plugin marketplace listed a tampered Anthropic Claude Code extension on October 29, version number 2.1.313, for about 11 hours, with as many as 5,000 developers potentially downloading it. Cursor has confirmed that the malware attempted to install a persistent SSH backdoor, collected usernames, hostnames, public IP addresses, Git email addresses, and the names of environment variables related to sensitive credentials and sent them to the attacker's server, and also added an attacker-controlled SSH key to achieve persistence. Cursor has removed the extension and temporarily shut down the entire extension marketplace to review other plugins and rotate access keys, while coordinating with the hosting provider to shut down the attacker's server, but it has not yet explained how the attacker published the tampered software package under Anthropic's name.

    Same story · 2 reports

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

  9. English text is being prepared.
    AI score53

    Anthropic restricts Opus 5.5 in Claude Code from replicating existing games

    Anthropic has restricted the Opus 5.5 model in its AI coding tool Claude Code from helping users replicate specific existing games. Developer @chasmmmmmmmmmmm said that around October 27, Claude Code added a rule that cannot be disabled: users may create original games, games of a certain genre, and games based on their own designs and assets, but may not replicate a specific existing game, even if renamed or claimed to be original. Anthropic technical staff member Addy Osmani responded that two lines of guidance have been added to the Claude Code prompt, and emphasized that using Claude to infringe intellectual property violates the platform's usage policy; currently, this restriction explicitly targets only Claude Code in the Opus 5.5 model.

    Same story · 4 reports

    #Product updates#Anthropic#English text is being prepared.

  10. English text is being prepared.
    AI score67

    OpenAI's latest funding target shrinks by $10 billion, plans to raise $20 billion at a valuation of about $1.25 trillion

    According to the Financial Times, OpenAI has lowered its latest funding round target from $30 billion by about $10 billion and is now seeking to raise about $20 billion at a valuation of about $1.25 trillion. The company previously planned to raise $30 billion at a valuation of $1.4 trillion and positioned the round as the "last chance" for investors to buy in before next year's IPO. People familiar with the matter said that because some investors reduced their subscription amounts, the first closing amount may be less than $20 billion; a consortium led by MGX is expected to be the largest contributor, while Sequoia Capital, Founders Fund, Coatue, Andreessen Horowitz and SoftBank also plan to participate.

    Same story · 7 reports

    #Industry news#OpenAI#Anthropic

  11. heise online
    AI score51

    US lawyers say Apple Watch's Audio Intelligence features may be illegal

    US lawyers warn that the Live Rewind and Siri Recap features on Apple Watch Series 12 and Ultra 4 may violate laws in some states that require all parties to consent to recording. The two features were announced in September and are not yet enabled; Apple says they will launch this year in Beta in English outside the EU. Live Rewind transcribes the most recent 15 seconds of speech, while Siri Recap automatically summarizes the day's meetings without distinguishing between speakers; Apple stresses that audio files are not stored on the device or in the cloud, but lawyers argue there is no legal precedent on whether transcription and summarization count as regulated recording, and expect litigation soon.

    #English text is being prepared.#Speech#English text is being prepared.

  12. Yicai
    AI score56

    Shanghai healthcare data moves from aggregation to operations, with three types of scenarios—real-world research, health insurance, and clinical agents—reaching scale

    Shanghai has built a health data platform that collects and integrates more than 140 billion records from over 400 major public medical and health institutions across the city, and has cumulatively completed out-of-domain delivery of 18 data products in the commercial insurance sector. Liu Xiao, head of data operations at Shanghai Data Resources Operations Co., Ltd., said that three types of scenarios—real-world research, health insurance restructuring, and clinical agents—already meet the conditions for large-scale development and utilization, and that development follows a model in which data does not leave its domain and only scoring reports, labels, and interfaces are delivered. The current bottlenecks have shifted to who supplies the data, who gets a share of the revenue, and how to sustain it. Shanghai's next step will be to clarify the revenue-sharing mechanism, establish a due-diligence exemption framework, and promote scenario standardization.

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

  13. The Decoder
    AI score75

    Microsoft offers voluntary departure packages to about 60,000 employees, with data center and AI-related teams excluded

    Microsoft is offering voluntary departure packages to about 60,000 employees, roughly a quarter of its total workforce, covering vice president-level and below employees in the United States, Canada, India and the United Kingdom. Xbox, Windows, Copilot and the finance department are all included, while data centers, sales, cybersecurity and quantum computing are excluded. This is Microsoft's second such program in 2026; in May it offered buyouts to 8,750 U.S. employees, and in July it laid off 4,800 people; total headcount remains about 223,000, and CFO Amy Hood said the company plans to invest about $170 billion in AI infrastructure in 2026.

    #Industry news#OpenAI#Microsoft

  14. Rohan Paul
    AI score60

    Sam Altman Meets About 50 Samsung Executives in South Korea, Seeking to Deepen Memory Chip and Data Center Cooperation

    According to Bloomberg, Sam Altman met with about 50 Samsung executives in South Korea, seeking to deepen cooperation on memory chips and data centers. The talks continue previous agreements under which Samsung supplies memory for OpenAI's Stargate project and collaborates with it to build data centers, including offshore floating sites. When the agreement was signed in October 2025, OpenAI said Samsung and SK Hynix together aimed to supply up to 900,000 DRAM wafers per month.

    Same story · 3 reports

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

  15. The Decoder
    AI score71

    OpenAI's latest funding round raised only $11 billion, below its $30 billion target

    OpenAI's latest funding round has so far raised only about $11 billion, less than half of its $30 billion target, with a valuation of $1.25 trillion, below its expected $1.4 trillion. According to The Wall Street Journal, the shortfall stems from waning interest among large investors, against the backdrop of a tech stock selloff triggered by weak quarterly results from Google, Amazon and Intel. OpenAI plans to turn to family offices and high-net-worth individuals to raise up to another $20 billion, and is reportedly planning to launch an IPO early next year; rival Anthropic has recently seen faster revenue growth and could go public as early as November.

    Same story · 7 reports

    #Industry news#OpenAI#Anthropic

  16. Rohan Paul
    AI score74

    OpenAI's funding round shrinks from $30 billion to about $20 billion, valuation target lowered to $1.25 trillion

    According to Bloomberg, OpenAI's current funding round has been reduced from $30 billion to about $20 billion, a 33% decrease, while the valuation target has been lowered from $1.4 trillion to $1.25 trillion, a reduction of only about 11%. According to the Financial Times, investors first objected to the price in late October and then cut the amount they were willing to invest. Even at $1.25 trillion before the new capital comes in, OpenAI's valuation is still about 47% higher than the $852 billion at the time of its $122 billion funding round in March. This pullback coincides with a 3.5% decline in the Nasdaq 100 index, as OpenAI's private fundraising becomes more closely tied to public-market sentiment on AI spending.

    Same story · 7 reports

    #Industry news#OpenAI

  17. cnBeta
    AI score71

    OpenAI has lowered its latest funding target from $30 billion to about $20 billion

    OpenAI has lowered its latest funding target from $30 billion to about $20 billion, and has so far secured about $11 billion in investment commitments, less than half of the original target. Last week, the company first lowered its proposed valuation from $1.4 trillion to $1.25 trillion, with the deadline for the first batch of investment commitments set for October 30. Reports say some institutions have reduced their subscription amounts, while MGX, SoftBank, Sequoia, a16z and others are still in talks to participate, and the deal terms also include an option to lease newly built computing capacity in the UAE.

    Same story · 7 reports

    #Industry news#OpenAI#Anthropic

  18. Phoenix Technology
    AI score65

    Concerns over AI investment returns intensify; Nasdaq 100 market value may evaporate over $2 trillion in October

    According to Reuters, the Nasdaq 100's market value could evaporate by more than $2 trillion in October, with declines exceeding 5.8% since the start of October, as investors' doubts deepen over the return cycle of AI infrastructure investment. Companies such as Meta, Alphabet, and Amazon continue to raise their AI spending forecasts, and Meta plans to raise funds through a record $97.5 billion financing, triggering concerns about equity dilution. Morgan Stanley expects global AI infrastructure capital expenditures to exceed $1 trillion next year, while JPMorgan expects global AI capital expenditures to reach $5.5 trillion by 2030.

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

  19. cnBeta
    AI score51

    Domestic AI comic dramas attract over 150 million overseas users, with production costs as low as 350 yuan per minute

    Domestic AI comic dramas have attracted over 150 million users overseas, with production costs as low as 350 yuan per minute, and the production cycle for one comic drama has been compressed from 90 days to about 10 days. The cost of AI translation and production per title has dropped to 300 to 500 yuan, more than 80% lower than manual translation; domestic short dramas mostly earn 1 yuan to several dozen yuan per 10,000 views, while overseas they can earn 1 to 10 US dollars. Institutions predict that the overseas micro-short drama market is expected to reach 6 billion US dollars in 2026, of which AI comic dramas will exceed 4 billion US dollars, an increase of nearly 390% year-on-year, with Chinese manufacturers holding a market share of over 90%.

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

  20. Analytics India Magazine
    AI score88

    Anthropic Files for IPO: Every $1 of Compute Spending Generates $2.08 in Revenue

    Anthropic has filed for an IPO in the United States and plans to list on the Nasdaq Global Select Market under the ticker ANT. The filing shows revenue of approximately $16.3 billion in the first half of 2026, with net loss widening to $80.7 billion, including $72.4 billion in non-operating accounting charges related to the fair value of convertible notes; in the second quarter of 2026, every $1 of compute spending generated $2.08 in revenue, up from $0.40 in the first quarter of 2025. The company's total compute and infrastructure commitments are approximately $637 billion, including at least $114.5 billion with Google, at least $110 billion with Amazon, and at least $44.9 billion with Microsoft, plus approximately $161.2 billion in equipment leases with Broadcom.

    Same story · 127 reports

    #Industry news#OpenAI#Anthropic