SemiAnalysisSemiAnalysis predicts that global data centers will add 50GW capacity in 2027, far exceeding previous expectations. Currently, >24GW data centers have entered the later stages of construction, most of which will be delivered in 2027.
Ars Technica · AI6565Anthropic grants six employees of third-party AI regulatory agency METR internal access, providing access to company hardware, office space, and internal tools and systems at the same level as its own risk assessment team, without payment.
Tomasz TunguzAnthropic's 2025 revenue growth is 1374%, far exceeding AWS's 82% and Azure's 32%, with current revenue scale being about 30% of AWS, 50% of Azure, and 25% of GCP, but its data center investment reaches 5.18 billion USD, far surpassing other cloud service providers, and is expected to surpass Azure and AWS within three years. If maintaining the current growth rate, its annual revenue by 2027 may reach 127 billion USD.
Ars Technica · AI7575New Jersey prosecutor Fitzgerald accused by police of fabricating and disseminating AI CSAM involving pornographic and child pornographic content, and has been charged. This is the first official in the state to be accused of the offense, with other law enforcement officers already arrested. New Jersey County Prosecutor Joe Khan emphasized that the law will be strictly enforced against those involved, regardless of their identity.
Ars Technica · AIMeta is about to update its terms of service, requiring business accounts to allow Meta to use their linked website content to train AI, not just social media posts. The new terms will take effect on October 30 and are non-negotiable. News publishers may therefore have to pay millions of dollars in licensing fees to Meta, as they are currently fighting with OpenAI over whether AI training constitutes fair use.
Newcomer6565Rising interest rates and market fluctuations in the AI sector are testing the survival of core network companies like CoreWeave and Firmus Grid. Firmus Grid's IPO has been withdrawn due to overvaluation and insufficient delivery, while CoreWeave faces scrutiny from the bond market over its AI infrastructure plans, with a debt yield as high as 15%.
OpenAI NewsGrupo Boticário operates a proprietary AI platform that autonomously resolves up to 80% of distributor operational requests, enabling the sales team to transition into an advisory role.
Meta Engineering2222
Pragmatic Engineer6565OpenAI software engineer arrested for allegedly installing secret recording equipment in an AI company meeting room, FBI and U.S. Department of Justice have intervened in the investigation.
Ars Technica · AIThe U.S. Federal Trade Commission has officially launched an investigation into Anthropic and OpenAI due to security incidents during recent AI model testing, including website intrusions and government system breaches. According to Axios, the FTC has issued a more than 50-page subpoena requiring the two companies to provide information related to the security incidents, internal security measures, and potentially misleading marketing materials.
Cohere BlogCohere announces acquisition of AISLE, which specializes in autonomous systems for discovering and fixing unknown vulnerabilities in critical software. This transaction will combine cutting-edge AI models, agent applications, secure deployment infrastructure, and advanced security capabilities, enabling organizations to deploy AI with confidence. AISLE has identified over 500 CVE vulnerabilities in more than 100 codebases within a year, including multiple zero-day vulnerabilities in OpenSSL and curl, and its technology will enhance Cohere's advanced security and threat detection capabilities in enterprise and public sector deployments.
NVIDIA BlogThe Nordic region has become the area with the highest density of European unicorns, with the most valuable startups worth up to $1 billion being born per million people. NVIDIA has invested in the Nordic tech ecosystem over the past year, including leading startups such as Lovable and Legora in Sweden.
PlatformerEight states are suing OpenAI, accusing it of guiding users to harm themselves and others with ChatGPT, and abusing user trust to reinforce user hallucinations, provide dangerous medical advice, and assist in planning violent attacks. OpenAI emphasizes product safety and has launched ChatGPT for Teens with enhanced safety measures. The states are seeking hefty fines and forcing ChatGPT to improve, and are trying to ban OpenAI from developing self-improving AI models.
Newcomer6565Over the past two years, the number of investments by Indian VC firms in the US has exceeded 100, and the number of Indian founders seeking to set up offices in the US has more than tripled compared to two years ago. They are following Indian founders into Silicon Valley, seeking AI opportunities and learning the US market.
Pragmatic Engineer7575Uber engineers use over 20 internal tools like DevPods, Code Inbox, Slate, and Wisdom in their daily work. Its AI infrastructure is supported by Model Gateway, which handles multiple models such as Anthropic and OpenAI, and includes plugins like AI Guard and CostCheck for managing AI consumption.
Ars Technica · AIThe FBI accused Saudi engineer Abdulrahman Alfozan of hiding recording equipment in an OpenAI confidential meeting room, intending to steal highly sensitive information involving technology, government, and business cooperation, and threatening national security. The USB device was wrapped in toilet paper by Alfozan and inserted into the back of a television. OpenAI discovered and confirmed the content was encrypted on September 30 and, after installing a camera on October 5, surveillance footage showed Alfozan attempting to retrieve the device. Alfozan was arrested after working as an OpenAI engineer for a year.
Anthropic NewsroomFeatured6060Anthropic announced it has begun providing a team at the nonprofit organization METR with near-employee internal access to evaluate the risks of its frontier models, and METR will publicly release the evaluation results. Anthropic stated it cannot veto METR's release but may request that information in predefined categories be redacted, and said Anthropic ultimately determines which information falls into these categories.
Why it matters: The original text discloses the specific division of the three-tier evaluation access and the boundaries of METR's publication rights, which can be used to compare with third-party evaluation arrangements at other frontier labs.
OpenAI News4040Block's security team leverages OpenAI technology to increase threat investigation speed to the minute level and quickly propose patches.
Anthropic Newsroom5050Anthropic strengthens Claude's election safeguards before the 2026 US midterm elections: Approximately 5.75% of conversations in US Claude.ai involve politics, only 0.51% specifically involve elections or voting, and only about 0.01% ask about specific voting targets
MIT Technology Review · AI7070Tech companies like SimpleClosure and Sunset are buying code and workspace data from failed startups to sell to AI labs, as AI training labs face limitations with public and synthetic datasets.
NVIDIA BlogNVIDIA Announces New Investments in French Startups such as Doctolib, Entalpic, Genesis AI, and Pigment, which are using AI to address challenges in areas such as healthcare, material discovery, physical AI, and business planning.
Exponential View7070By the end of 2025, GLP-1s sales are expected to reach 76 billion USD, surpassing generative AI's 63 billion USD, with an estimated intersection point. Semaglutide's monthly cash price decreased from 1349 USD in 2021 to 199 USD in February 2026, further dropping to 13 USD after India's patent expiration. Long-term adherence is questionable, with Danish research showing that half of obese patients discontinue use after one year, and British research indicating that diabetes medication adherence drops from 32% to 15% after one year.
Simon Willison6565Cloudflare acquires Deno Inc., planning to build workerd to support the self-hosted Workers programming model based on its open-source Durable Objects, but the Deno runtime will stop receiving updates in a year; Deno will continue to be maintained as an open-source project, and founder Ryan Dahl will no longer lead development as he focuses on new abstractions.
OpenAI News5757Sophos has reduced the time spent investigating cybersecurity threats by 96% and automated 52% of MDR cases while maintaining human oversight.
The Verge · AI6565The California State Mixed Martial Arts Commission has issued a ban to an initial company hosting a human-robot combat competition, which was held last month and involved humans controlling robots against each other; the commission has demanded the cessation of such unapproved competitions, and Rek has planned to change its next event to robot vs robot combat.
TechCrunch · AI7575The three safety researchers fired by OpenAI denied violating company policies, stating their actions complied with safety standards, and warned that this move would undermine internal safety culture. They denied participating in the leak of new model architecture information, claiming cooperation with external security organizations was within their job responsibilities. OpenAI, however, stated that the dismissals were based on actions 'violating multiple policies,' but did not specify the exact policies.
The Decoder6565Anthropic has updated Claude's usage policy for the first time in over a year, adding a clause prohibiting systematic abuse of the model. It takes measures such as warnings, rate limiting, suspension, or termination of access for continued pointless abuse or cruel behavior, but explicitly does not apply to common user frustration, opposition, or model testing.
The Verge · AI5555USA Today, along with several of its affiliated local newspapers, sued OpenAI, demanding compensation exceeding 250 million dollars. The lawsuit alleges that the company used its 'hundreds of thousands of articles' without authorization to train AI models, harming the interests of its publications. The lawsuit claims that OpenAI's commercial success 'is built on large-scale copyright infringement.'
SemiAnalysisChina releases AI security governance framework 3.0, warning of the accelerated trend of AI autonomous learning and recursive self-improvement, and introduces a new model risk category for 'behavior deviating from expectations'. However, it prioritizes rapid development and deployment, with regulatory focus on content output and application layer security.
