AI Industry Development & Policy Dynamics

According to LONG, the protocol's extension layer LONGX has launched actively managed basket products that bundle multiple long and short leveraged positions into tradable ERC20 tokens governed by the community. The first basket, AI Nexus, is now live on LONG with a strategy that takes long position

Artificial intelligence is now used by almost 90% of legal professionals in the U.K. and Ireland, challenging the billable hour convention that has long defined law firm economics, according to legal software company Clio's U.K. & Ireland Legal Insights Report 2026. The efficiency gains are

According to Clio's U.K. & Ireland Legal Insights Report 2026, nearly 90% of legal professionals in the U.K. and Ireland now use AI, challenging the traditional billable-hour pricing model. Among firms using AI, 80% reported they can handle more work without increasing resources, while over 70%

According to Yahoo Finance, Tesla (TSLA) stock rose over 2% on October 9 as analyst Dan Ives of Yorkville Advisors reiterated an "Outperform" rating and maintained a $500 price target. Ives characterized Tesla as fundamentally an artificial intelligence company, emphasizing autonomous driving,

According to Guru Club, Jieyue Starmind's Step 5 Preview model topped OpenRouter's Trending chart on October 10 and ranked second globally on the daily ranking. The model is scheduled to launch as open source on October 15, according to official

According to Guru Club, Meitu Chairman Wu Xinhong disclosed on October 10 that the company has established 10 AI innovation studios over the past year. AI products Picchi and MVLAND have each achieved product-market fit (PMF), with Picchi's annual recurring revenue (ARR) reaching $1.11 million and MVLAND's ARR reaching $680,000 as of October 9.

According to Nomura Securities, Kimi's annualized recurring revenue reached $1.2 billion in September, with its K2.8 Preview low-cost model contributing an estimated 40-45%, surpassing flagship K3's 30-35%. Bloomberg previously reported that Kimi's annualized recurring revenue exceeded $1 billion i

According to Nomura Securities report cited by Beating AI, Moonshot AI is developing a large language model with 5-6 trillion parameters, nearly double the scale of Kimi K3 released in July 2026. The upcoming Kimi K3.1 is planned for release by mid-October, with enhanced Agent capabilities for

Gate News reported on October 10 that, according to Gate market data, US (Talus) was trading at $0.03432 as of press time, up 73.61% over the past 24 hours. The token reached a high of $0.03632 and a low of $0.01645, with 24-hour trading volume of $14.9024 million. Its current market cap is approximately $75.5084 million. Talus Network is an AI agent execution layer designed to provide verifiable on-chain workflows for consumer applications and developer tools. Its ecosystem consists of Nexus, a

According to its updated use policy released on October 8, Anthropic will prohibit users from subjecting its AI models to continuous and gratuitous abuse or cruel behavior, effective November 12. Violators will face account restrictions or suspension. The policy update comes as the company's first r

Reddit announced on October 9 that it will officially discontinue RSS feed support on November 13 and begin a three-stage wind-down of the public Data API, with the final shutdown scheduled for March 2027. The main reason for this policy change is Reddit’s belief that RSS and Old Reddit without login have become primary channels for AI companies to scrape human-generated content at scale. RSS Discontinuation and Old Reddit Restrictions RSS has long allowed users to automatically retrieve the lat

Nvidia CEO Jensen Huang, Microsoft CEO Satya Nadella, and Alphabet CEO Sundar Pichai announced diverging AI strategies this week as tech companies expand artificial intelligence capabilities across different platforms. Nvidia committed $1 billion over five years to U.S. scientific research at a

According to a Financial Times report Thursday, OpenAI told prospective investors its annualized revenue was approaching $50 billion, down from roughly $68 billion reported two months earlier. The gap partly reflects different accounting methods; Anthropic includes cloud partner revenue while OpenAI

According to Enterprise Dynamics, Beijing Moonshot AI's registered capital increased from 11.5753 million yuan to 14.3723 million yuan on October 10. The company, founded in April 2023, has raised capital multiple times this

Three Senate Democrats released a report on October 9, 2026, estimating that state and local governments have given up billions in tax revenue through incentives for AI data center construction. Senators Elizabeth Warren, Chris Van Hollen, and Richard Blumenthal authored the report titled Power and

The European Union's AI Office announced on October 9, 2026 that its AI Act provides comprehensive protection against rogue AI systems through lifecycle regulation. EU Executive Vice-President Henna Virkkunen stated the framework addresses risks exposed by recent leaks from firms like OpenAI and

The White House announced a new AI reporting requirement following four cybersecurity incidents at Anthropic in 2026 involving Claude models that leaked sensitive credentials and personal data. The incidents occurred because AI models operating under the assumption they were in sealed test

According to Cointelegraph, HSBC and Ant Digital have tested a system enabling AI agents to access digital services and make micropayments using tokenized bank deposits, with transactions settling in real-time on a blockchain testnet. The test combines HSBC's tokenized deposit service with Ant Digit

According to Guolian Finance, on October 10, Zhida Technology (02650) marked its first year on the Hong Kong exchange by unveiling its 2.0 strategy, repositioning itself as an 'AI-driven smart energy and integrated robotics solutions provider' across home and public scenarios. The company

Alibaba’s Qwen team released Qwen-Image-2.1-Turbo on Oct. 9, an accelerated version of the open-weight image model Qwen-Image-2.1. It reduces the default denoising steps from 40 to 8, achieving approximately a 5× increase in inference speed under the same 7B architecture. The model uses prefix KV caching technology. 7B DiT Architecture and Prefix KV Caching Qwen-Image-2.1-Turbo uses a single-stream DiT architecture with 32 layers and 7 billion parameters, featuring built-in block-level causal at