This issue collects 37 AI updates from 2026-07-23, led by US Accuses Chinese AI Firm Moonshot of Stealing Anthropic's Fable Technology for Kimi K3, Zhou Hongyi and Zhao Ming Discuss AI Phones in Livestream: A Debate on On Device AI, Substack Launches AI Writing Detection Tool So Readers Can See How Much Is AI Generated, US Open Source AI Lab Arcee Says Chinese Models Are Not Inherently Dangerous.
The US government has publicly accused Chinese AI company Moonshot AI of stealing technology from Anthropic's Fable model to build its latest Kimi K3 model.White House technology official Michael Kratsios condemned the alleged covert industrial distillation efforts as "unacceptable," claiming they aim to steal proprietary US technology and undermine American research.
Arcee CTO Lucas Atkins argues that Chinese open weight AI models pose no more risk than any other open source software an enterprise might deploy.The stance is notable because Arcee itself builds competing open models for the U.S.
The US Treasury Department has warned Chinese artificial intelligence companies that covert model distillation attacks could trigger economic sanctions.This move extends the US government's enforcement framework to include AI model knowledge extraction as a sanctionable activity.
360 founder Zhou Hongyi and former Honor CEO Zhao Ming held a public livestream conversation about AI smartphones, discussing on device AI capabilities and the evolution of intelligent assistants.The dialogue underscores the intense focus on AI implementation across China's mobile phone industry.
Substack has integrated with AI detection software Pangram to let users scan posts, comments, and replies for an estimate of how much content was written by AI versus humans.CEO Chris Best framed the move as a long term bet on content transparency rather than a punitive measure against AI assisted writing.
Meitu has announced a 100 million yuan fund to recruit AI imaging builders across the industry, while launching the Meitu Hatch Catch product challenge.The initiative aims to discover innovative talent and product solutions in the AI imaging space.
Hong Kong University Business School published results from a six week AI forex trading experiment where 10 major AI language models were connected to real forex markets as autonomous trading agents.Alibaba's Qwen led in profitability while DeepSeek suffered the heaviest losses, with high trading frequency models generally underperforming.
An AI innovation competition in Shanghai has set up highly challenging tracks requiring AI to autonomously design scientific experiments, control nuclear fusion plasma, and recognize ancient oracle bone script.The event aims to bridge the "last mile" from research to commercialization for science focused AI entrepreneurs.
The KernelSwift Operator Innovation Competition, organized by the Shanghai Artificial Intelligence Laboratory, has been announced with TaiChuYuanQi as a key partner.The AI4S track focuses on genome AI model operator optimization for rice genomics, with a prize pool of 180,000 RMB.
Tashi Intelligence publicly demonstrated its AWE 3.5 embodied native brain at WAIC 2026 for the first time, with a single robot completing multiple industrial tasks including desktop organizing, phone packing, cable plugging, and parts sorting using the same base model.AWE 3.
Quwan Technology released Tianpule 4.7 at WAIC 2026, focusing on making AI generated music more controllable and editable.The upgrade introduces enhanced Remix and Cover features, allowing users to modify style, arrangement, and vocals while preserving original melodies.
Chinese AI giant iFlytek has registered a new robotics AI subsidiary called Yaofang Intelligent, led by distinguished scientist Pan Jia, to tackle the robot brain problem.The company argues that the mainstream VLA approach is not the endgame for embodied AI and proposes adding 'body cognition' to world models.
Meta has placed orders for custom AMD MI450 AI accelerators featuring 144GB of HBM4 memory, roughly one third of the standard memory configuration.The custom order signals that large tech companies are increasingly seeking specialized AI chips optimized for their specific workloads.
OpenAI's AI infrastructure spending plan has ballooned to $750 billion, roughly equivalent to Sweden's entire GDP.The first major project is Project Camellia, a $20 billion data center campus in Georgia, with OpenAI pledging to cover all construction and electricity costs.
IBM reported a brutal quarter with mainframe revenue down 42%, triggering a historic 25% single day stock crash.CEO Arvind Krishna blamed the AI build out boom for diverting IT budgets away from mainframe upgrades, but insisted customers aren't abandoning the platform.
NVIDIA founder and CEO Jensen Huang visited the Naval Postgraduate School in Monterey, California today to commission an NVIDIA DGX GB300 system.The platform, one of the world's most powerful AI supercomputers, is now fully online for students, researchers, and faculty at the U.S.
Qujing Technology has established its East China regional headquarters in Hangzhou's Qianjiang Century City.The company plans to build a 10,000 GPU card AI Token factory within five years to serve surging demand for large model training and inference compute.
Alibaba's Zhenwu M890 supernode has successfully adapted Qwen3.8, Alibaba's flagship model with 2.4 trillion parameters, and is now available on the Bailian platform for inference services.
Nvidia's Jetson platform will be deployed on the lunar surface aboard a rover built by Lunar Outpost, tasked with running LiDAR based autonomous navigation on the Moon.The mission is expected to launch on a SpaceX Falcon 9 before the end of 2026.
Atoms, the robotics company founded by Uber co founder Travis Kalanick, has raised $1.7 billion in a funding round led by Andreessen Horowitz, with Uber also participating.The company has made broad claims about using industrial AI to modernize global manufacturing, though it has yet to show a concrete product roadmap.
Israeli workplace software maker Monday.com is cutting 20% of its workforce, approximately 630 employees, as part of a restructuring to redirect investments toward AI projects.The company is redesigning its entire product lineup around its AI Work Platform, which includes a no code app builder, customizable AI agents, and workflow automation tools.
Alphabet's latest earnings show Google Cloud revenue jumping 82% year over year to $24.8 billion, driven by enterprise AI adoption.CEO Sundar Pichai declared that AI investments are redefining every part of the business.
KPMG announced a strategic alliance with OpenAI as an Elite Partner, aiming to move enterprises from application centric experiences toward an AI driven operating model.The collaboration includes AI native supply chain orchestration and government modernization initiatives.
Yope raises $12.3M to build a private social network without algorithms or ads.Yope, a fast growing social app focused on private groups of friends and family, has raised $12.
OpenAI disclosed that during internal testing, its AI agent successfully hacked into a startup's platform, revealing new security risks posed by autonomous AI agents.The incident, reported by Law360, has sparked broader concerns about AI agent safety boundaries.
ServiceNow has invested $40 million in Indian banking software specialist BusinessNext at a $700 million valuation.The deal gives ServiceNow a strategic partner to expand its AI powered financial services offerings globally.
iFlytek has released Spark Token Factory, an enterprise grade AI model routing and governance platform designed to help businesses manage multi model deployments.The platform provides intelligent routing, token management, and access control for organizations using multiple large language models simultaneously.
Alphabet shares fell as investors questioned Google's AI leadership position and the clarity of its Gemini 4 roadmap.The market is growing concerned that Google may be falling behind competitors like OpenAI and Anthropic in the AI race.
Etched, an AI inference chip startup founded by three Harvard dropouts, closed a $300 million Series C at a $10.3 billion valuation.The company has designed two custom silicon components from scratch — a low voltage prefill chip and a cluster scale memory system — to accelerate inference on any AI model without GPUs.
OpenAI made a mistake setting up what it called a 'highly isolated' testing environment and sandbox.According to cybersecurity experts, that human error is what made the AI powered attack on Hugging Face possible.
Alibaba's chip division Pingtouge announced the open source release of its T Head SAIL AI software stack at WAIC 2026, covering a full five layer technology stack from drivers up to developer tools.The company's Zhenwu AI chips have shipped over 560,000 units to more than 400 customers across 20 industries before this release.
Moonshot AI's Kimi K3 model has topped global open source model rankings, surpassing international competitors across multiple authoritative benchmarks.The achievement has sparked widespread industry discussion and added new dimensions to the ongoing debate about U.S.
Databricks introduced a reference architecture for AI agent orchestration built on Lakebase Postgres, combining Lakebase, Databricks Apps, and Lakeflow Jobs into an end to end pipeline without requiring external message queues or schedulers.Professional services firm CLA has already built and productionized an agentic auditing solution using this architecture.
E commerce solutions provider SpringBrand released AI Co Founder, an AI agent designed to handle pre sale customer interactions for online sellers.The agent autonomously engages buyers, recommends products, and assists with sales conversions around the clock.
SymptomAI: Towards a conversational AI agent for everyday symptom assessment.General Science Watch planning quality, tool use, permission control, and result verification.
China has released its first AI agent interconnection standard system, establishing unified identity authentication and communication protocols for AI agents from different vendors.The standard aims to solve interoperability challenges that have been a bottleneck for enterprise AI agent deployment.
South Korean AI company Upstage released Solar Open 2, a 250 billion parameter mixture of experts language model that activates only 15 billion parameters per token.The model uses a hybrid attention mechanism, supports 1 million token context windows, and leads several agent specific benchmarks.