All news

MMediaNama20:32

AI score58

A US court is hearing whether training AI on song lyrics constitutes fair use, while India plans to introduce compulsory licensing

Universal Music Group, Concord and ABKCO sued Anthropic for training Claude on the lyrics of hundreds of songs; both sides have filed for summary judgment. Anthropic argues that the training constitutes fair use, while the publishers say users sent Claude nearly 120,000 requests to generate synthesized lyrics within six months. Two judges in the Northern District of California reached different conclusions in 2025 on whether AI training constitutes fair use, and Anthropic also faces lyrics-related lawsuits from Sony Music Publishing and Round Hill Music. MediaNama has asked Anthropic, the publishers' lawyers and IMI about whether Anthropic obtained lyric licenses from Indian rights holders, whether Claude's lyric anti-copying measures cover songs in Indian languages, and the royalty issues surrounding India's proposed licensing.

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

TThe Elec20:32

AI score62

Nvidia prepays six suppliers including Samsung Electro-Mechanics and Ibiden to expand production, locking in AI infrastructure capacity

On October 21, Nvidia announced partnerships with Samsung Electro-Mechanics, Ibiden, Amkor, KYEC, SPIL, and Unimicron to lock in packaging substrate, testing, and co-packaged optics (CPO) capacity through prepayments. Among these, the agreements with Ibiden, Samsung Electro-Mechanics, and Unimicron will expand semiconductor packaging substrate capacity 2.5-fold over the next three years; Ibiden's Ono plant in Japan will triple its capacity and reach full production in the first quarter of 2028. Nvidia also prepaid $1.1 billion to KYEC, doubling its total testing capacity in 2027 compared with 2026, and is working with SPIL to increase CPO optical engine capacity 15-fold by mid-2027.

Same story · 7 reports

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

TTechCrunch AI20:06

AI score75

Government websites worldwide go offline due to a PHP vulnerability, while Node.js, LiteLLM and others disclose security issues on the same day

After AI discovered a critical security vulnerability in PHP, government websites in many countries around the world went offline, and the U.S. National Security Agency recommended that organizations using PHP update immediately or temporarily take their systems offline. On the same day, Node.js fixed four high-severity vulnerabilities, one of which allows remote code execution; the open-source LLM request proxy LiteLLM was hit by attackers who injected malicious code, and the malicious versions were published to PyPI, Docker Hub and GHCR, containing an information stealer that harvests cloud credentials, passwords and AI service credentials as well as a remote access trojan. LiteLLM said it has removed the malicious versions and images. In addition, the private key of the Indian certificate authority eMudhra was compromised, and security professionals say the cleanup effort will be larger than that of Heartbleed.

#Safety / alignment#Open-source ecosystem#English text is being prepared.

TThe Rundown AI20:04

AI score62

OpenAI's GPT-5.6 Sol helped discover two critical PHP vulnerabilities

OpenAI's GPT-5.6 Sol helped security firm Calif and Google-owned Wiz discover two critical vulnerabilities in PHP. PHP has released patches for them, and the U.S. National Security Agency recommends that those unable to patch immediately take their systems offline. The first vulnerability was discovered by Calif on August 1 with the help of 5.6 Sol, and Wiz independently discovered both vulnerabilities weeks later using Sol and Gemini 4 Argon; OpenAI said it worked with PHP maintainers for two months to fix them, and called this one of the most impactful security issues in recent years. These two vulnerabilities, which have existed for about ten years, could allow attackers to hijack servers without logging in. The governments of Finland, Germany, and the Netherlands have taken some websites offline, and Cloudflare said early attacks have already occurred.

#Safety / alignment#OpenAI#Google

TThe Verge AI20:00

AI score62

Eiliko maker says it will hire an independent security firm, but the vulnerabilities remain unfixed

Energize Lab responded to The Verge regarding the Eiliko robot security vulnerabilities, saying it has confirmed the core issues in the report and is addressing them one by one, will hire an independent security firm to conduct a full review of its systems, and has set up a dedicated email address at security@energizelab.com. The company said it has closed all external access to its backend systems, disabled the interface that could remotely modify robot replies, stopped storing conversation records and user profiles, and will delete stored conversation records, user profiles, and avatar images. However, The Verge verified with security researcher Sammy Azdoufal that, as of the time the company sent its letter, customer recordings and photos were still stored at completely unprotected public URLs, and Azdoufal still retained full SSH access to these robots.

#Safety / alignment#Embodied AI#Industry news

EEast Money19:42

AI score66

MiniMax Breaks Issue Price for First Time in Hong Kong; Zhipu Falls 7.1% on Same Day as LLM Price War Continues

On October 22, MiniMax's Hong Kong shares fell more than 7% intraday, hitting a low of HK$158.10 and closing at HK$160.70, below the HK$165 issue price—its first break below the issue price since listing in January this year; on the same day, Zhipu closed down 7.1% at HK$460.80. MiniMax's share price has pulled back about 88% from its high, while Zhipu has pulled back about 85% from its high, with more than HK$1 trillion in market value wiped out. The article attributes the pullbacks to weakness in Hong Kong stocks and the price war: Zhipu launched GLM-5.3-Flash on August 26, pricing it at one-tenth of the flagship GLM-5.3; OpenAI released the GPT-6 series on September 23 and announced a permanent 50% API price cut; MiniMax released M3.1 Flash on October 19, saying the overall cost of use for Agentic tasks is 70% lower than the previous generation.

#Industry news#DeepSeek#OpenAI

CCNN Business Tech18:50

AI score93

AI-assisted discovery of critical PHP vulnerability PHP2Shell prompts emergency takedowns and patching of numerous websites worldwide

Teams from OpenAI, Wiz, and Calif used AI to discover two long-standing critical vulnerabilities in PHP, collectively called PHP2Shell, which could allow attackers to execute code remotely and read memory containing sensitive information such as passwords. PHP underpins 70% of websites worldwide, and the PHP Foundation has released patches; the U.S. NSA and European cybersecurity agencies advised enterprises to act immediately, while websites of Ireland's The Journal, Japan's ANA, JR West, MUFG, and government sites in France, Denmark, Finland, and the Netherlands were temporarily taken offline. Wiz said researchers used Google's Gemini 4 Argon, Wiz Atlas, and OpenAI's GPT-5.6 Sol, with a single researcher completing in one week work that previously took world-class teams months, and compared the incident to 2014's Heartbleed.

#Safety / alignment#OpenAI#Google

BBusiness Insider Tech18:49

AI score60

Anthropic engineer says Claude has written all of his team's code

Katelyn Lesse, Anthropic's head of platform engineering, said at Grafana Labs' ObservabilityCON event that her team of about 200 people has all of its code written by Claude, with humans responsible only for system design and verification. She said the team once migrated a large codebase from Python to Rust, with the build, monitoring and verification steps all handled by agents. Anthropic now allows agents to fix tests autonomously and is considering letting agents modify the codebase on their own. In May this year, Anthropic said Claude had written more than 80% of the company's production code.

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

TThe Beijing News18:10

AI score50

AI lands on the Double 11 shelves as large models start selling by the item

This Double 11, AI appears in two roles at once—as a merchant tool and as a consumer product. Large model vendors such as Zhipu and Qwen Office have opened official flagship stores on Tmall; Tmall has launched a token top-up section; and JD.com has upgraded its AI shopping assistant Dongdong and launched video shopping digital humans. The JD Logistics Super Brain large model 3.0 has achieved parcel volume forecast accuracy above 94.2% and an adoption rate above 90.2% for optimal delivery route recommendations, with more than 90,000 merchants using AI digital human livestreaming. Tmall packages tokens into monthly, quarterly, and annual cards for scenarios such as study, family, and solo studio use, targeting individual users and small teams with clear AI usage needs.

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

CCailian Press18:01

AI score65

OpenAI sends letter to California Governor Newsom, recommending strengthening of AI transparency bill SB53

In an October 20 letter to California Governor Newsom, Chan Park, OpenAI's Head of Policy and Partnerships for the U.S. and Canada, said that OpenAI supports imposing mandatory requirements on AI companies, including equipping models with emergency shutdown mechanisms, undergoing independent audits annually, implementing higher-standard cybersecurity norms, and recommended strengthening SB53, California's AI transparency bill that took effect in 2025. The letter proposed that regulatory rules should make clear that corporate safety frameworks must cover the ability of frontier models to autonomously conduct AI research and development. OpenAI had publicly opposed overly burdensome state-level legislation during the SB53 legislative process, and has now shifted its position.

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

PPhoenix Technology17:48

AI score61

Microsoft adds Guardian Agents to its security system Project Perception to supervise enterprise Agents

On October 19, Microsoft announced that it is adding a group of "supervisors" called Guardian Agents to its security system Project Perception, specifically to check the behavior of other Agents within the enterprise, and upon detecting anomalies, investigate the cause and revoke permissions or halt operations according to enterprise-preset rules. The solution requires first establishing an independent identity for the Agent performing the task, granting only the permissions needed for the current task, and revoking them after the task ends; during operation, permissions can be tightened, human approval can be required, or the Agent can be isolated and stopped. Project Perception is currently still in a preview stage for a small number of invited customers, and is billed by Security Compute Unit (SCU).

#Industry news#Agents#Safety / alignment

336Kr · AI17:37

AI score78

The first batch of mass-produced Tesla Optimus robots rolls off the line; Musk says reaching 20,000 units per week will take another year

In its business update report, Tesla confirmed for the first time that Optimus has begun production on a high-volume production line at the Fremont factory, and what rolled off the line is the first batch of products with mass-production quality. On the earnings call, Musk said that reaching a rate of 20,000 units per week and an annualized one million units will take at least another year or even longer; the external delivery timing is set around the end of 2027, and initially only industrial scenarios will be pursued. He also mentioned that the hand design has not yet been finalized, training data is accumulated through robots playing against themselves, AI5 is planned to be integrated into the robots in April next year, and its computing power is about an order of magnitude higher than AI4.

Same story · 11 reports

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

336Kr · AI17:37

AI score82

DeepSeek's hundred-billion financing nears completion, allies with industrial capital including Perfect World and SMIC Capital and bets on embodied intelligence

According to Bloomberg, DeepSeek's second funding round is locked at about 100 billion yuan, with a pre-money valuation of about 500 billion yuan; the fundraising target was raised twice from the initial 50 billion yuan and then doubled. Existing shareholders Tencent and CATL each added about 10 billion yuan; new entrants include industrial capital such as Shixi Capital, SMIC Capital, CPE Yuanfeng, Legend Capital and Geely. Perfect World, Century Huatong and Litong Electronics participate indirectly through two special funds, Lisi Xingshen and Shixiang Shangshi; DeepSeek is also preparing for a STAR Market IPO and has already invested in embodied intelligence companies such as Unitree.

Same story · 14 reports

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

336Kr · AI17:37

AI score58

Jensen Huang Unveils DSX POD AI Factory Design and OpenShell Agent Architecture at GTC Berlin

At the GTC conference in Berlin, Jensen Huang unveiled the DSX POD AI factory design, in which Vera Rubin racks, network cabling, and liquid cooling systems are pre-integrated at the factory and delivered in containers, cutting on-site deployment time from about 6 months to as fast as 14 days; partners including Dell, HPE, Lenovo, and Schneider Electric plan to launch related products in early 2027. Huang said Vera Rubin has entered full production and released the OpenShell reference architecture, which places Agents in isolated environments governed by permission policies and independently monitors their behavior through software running on BlueField-4. NVIDIA also updated the Isaac open robotics platform and, in partnership with Agile Robots, introduced the Agile ONE SN humanoid robot reference platform, planned to be available to researchers and developers in early 2027.

Same story · 20 reports

#Industry news#Agents#Embodied AI

336Kr · AI17:37

AI score65

Citi and Bank of America Restrict AI Agent Access to Customer Accounts

Citigroup and Bank of America have followed JPMorgan Chase in restricting AI Agent access to customer accounts. Citi prohibits third-party agents from logging in directly or executing transactions but retains read-only access authorized by customers, while Bank of America restricts agent access to consumer deposits until responsibility-allocation agreements for authorized payments are determined. On its October 13 third-quarter earnings call, JPMorgan Chase announced the ban on such access, citing the risk of thousands of malicious agents, while advancing a targeted small-scale pilot of its cash management AI tool Smart Cash. The article also mentions that Meta's Muse has approached 9 million global downloads just over a month after launch, and a Visa survey shows that only 23% of surveyed U.S. consumers trust generative AI to complete payments on their behalf directly.

#Industry news#Agents#OpenAI

SSemiconductor Industry Review17:36

AI score56

NVIDIA Partners with Six Companies to Expand Substrate, Advanced Packaging, and CPO Capacity

At GTC Berlin, NVIDIA announced coordinated capacity expansion with six supply chain partners: Amkor Technology, Ibiden, King Yuan Electronics, Samsung Electro-Mechanics, Siliconware Precision Industries, and Unimicron. The expansion covers substrates, advanced packaging, testing, and optical engines. Substrate capacity is planned to increase to 2.5 times the current level over the next three years, and optical engine capacity is targeted to reach 15 times the current level by mid-2027. NVIDIA will also provide King Yuan Electronics with a $1.1 billion capacity reservation prepayment to build a new testing facility in the United States.

Same story · 7 reports

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

VEnglish text is being prepared.17:36

AI score60

After cumulative losses of about $96.3 billion, Meta Reality Labs shifts its focus to AI glasses and personal superintelligence

Meta's financial report for the third quarter of fiscal year 2026 shows that the Reality Labs division had revenue of $443 million and an operating loss of $4.129 billion, with the loss narrower than market expectations, bringing cumulative losses since 2020 to about $96.3 billion. Meta CFO Susan Li said the narrowing loss mainly came from revenue growth in AI glasses, partially offsetting the decline in Quest headset sales. Zuckerberg said on the earnings call that the pace of VR hardware iteration will slow, personal superintelligence has become the focus of Reality Labs' business, MUSE will be integrated into Meta smart glasses starting next month, and paid subscriptions are expected to increase.

Same story · 9 reports

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

Hheise online17:35

AI score56

Tesla's third-quarter net profit falls 43%, Musk pushes Optimus humanoid robot

Tesla's GAAP net profit for the third quarter of 2026 fell about 43% year over year to $787 million, revenue of $2.805 billion was essentially flat compared with the same period last year, and operating profit fell 56% to $718 million. R&D spending rose from $1.63 billion to $2.76 billion, which Tesla said was partly for projects including "Super Intelligence." At the analyst call, Musk highlighted the Optimus humanoid robot, saying that the Fremont factory has begun early-stage mass production, but that reaching a pace of 20,000 units per week "may take at least a year," and he did not disclose weekly output, external customers, or per-unit cost.

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

MMediaNama17:32

AI score68

Meta Q3 Earnings Call: Over 10 Million People Set Up Muse Within Six Weeks, Coming to Meta Glasses in November

On its October 21 earnings call, Meta said that within six weeks of launch, more than 10 million people in the US and Canada had set up Muse, about 25% of users use it daily, shopping is the most common scenario, and the average order value exceeds $200. Zuckerberg said Muse will launch on Meta Glasses in November 2026, and the company also plans to launch a hardware device called Muse Charm. CFO Susan Li said that apart from subscriptions, Muse has not yet been monetized; Muse runs in an isolated virtual machine in the cloud, and its initiated operations are reviewed by the Sentinel system.

Same story · 11 reports

#Industry news#Agents#OpenAI

TThe Guardian Technology17:30

AI score60

Barclays to cut 5,000 jobs, management says AI boosts productivity

Barclays plans to cut about 5,000 full-time positions, roughly 5% of its 93,000 global employees, as part of a plan to save more than £1.1 billion by 2027; it has already cut 1,000 jobs this year. Chief Financial Officer Anna Cross said about 60% of the measures involve support functions, including reducing developers, because AI makes employees more efficient; CEO CS Venkatakrishnan called AI the icing on the cake of the bank's technology modernization, with efficiency gains coming mainly from technology investment and process simplification. The bank also reported pre-tax profit of £2.3 billion for July to September, up 12% year on year, and announced a £500 million share buyback.

Same story · 4 reports

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