Daily updated AI industry latest developments, application innovations, and cross-border marketing trends — helping you quickly grasp important news and key signals worth noting.
AI Industry Hot News
Anthropic's Project Glasswing discovers over 10,000 critical software vulnerabilities
Anthropic's Project Glasswing, a collaborative security initiative with about 50 partners including Amazon, Microsoft, Google, and Nvidia, has used Claude Mythos Preview to discover more than 10,000 high- or critical-severity vulnerabilities across critical software systems. Cloudflare alone found 2,000 bugs including 400 high/critical severity. The UK AI Security Institute reports Mythos Preview is the first model to solve both of their cyber ranges end to end.source：Anthropic
Trump postpones AI executive order on government oversight at last minute
The Trump administration postponed signing an executive order on government oversight and access to AI at the last minute on Thursday, May 21. Trump said the order 'could have been a blocker' for jobs and 'tremendous good' AI is creating. China was cited as a factor, with Trump stating 'We're leading China. We're leading everybody.' Officials are concerned about political repercussions if a devastating AI-enabled cyberattack occurs.source：The Verge
Microsoft phases out Claude Code licenses, pushes Copilot CLI internally
Microsoft is removing most of its Claude Code licenses for the Experiences + Devices team by June 30, pushing developers to use GitHub Copilot CLI instead. The decision comes as Claude Code became popular inside Microsoft but undermined Copilot CLI adoption. Sources say it is also a financial decision tied to the end of Microsoft's fiscal year. Microsoft still continues to favor Anthropic's Claude models inside Microsoft 365 apps and Copilot.source：The Verge
OpenAI named Leader in enterprise coding agents by Gartner
OpenAI has been recognized as a Leader in enterprise coding agents by Gartner, validating the company's position in the growing AI-powered software development market. The recognition comes as enterprise adoption of AI coding tools accelerates across industries.source：OpenAI
AI Innovation & Applications
Resolve AI launches multi-agent system for production debugging with 2x accuracy
Resolve AI announced a major platform expansion featuring a multi-agent investigation system that dispatches coordinated teams of specialized agents to diagnose production failures in parallel. The system independently verifies conclusions and constructs complete causal chains from root cause to symptom. The company claims more than twofold improvement in root cause accuracy compared to earlier versions. The system can triage within five minutes before human engineers get involved.source：VentureBeat
Cerebras runs trillion-parameter Kimi K2.6 model 6.7 times faster than GPU clouds
Cerebras Systems announced it is now running Kimi K2.6, a trillion-parameter open-weight model developed by Beijing-based Moonshot AI, at nearly 1,000 tokens per second for enterprise customers. Independently verified by Artificial Analysis, Cerebras achieved 981 output tokens per second, making it 6.7 times faster than the next-fastest GPU-based cloud provider. For a standard agentic coding request, Cerebras delivered the full response in 5.6 seconds versus 163.7 seconds on the official Kimi endpoint.source：VentureBeat
MOSS: Self-evolving autonomous agents rewrite their own source code
Researchers from arXiv presented MOSS, a self-evolving autonomous agent system that performs source-level adaptation on production agentic substrates. Unlike existing systems confined to text-mutable artifacts, MOSS can rewrite its own agent harness code. The system uses a deterministic multi-stage pipeline with automatically curated production-failure evidence. On OpenClaw, MOSS lifted a four-task mean grader score from 0.25 to 0.61 in a single cycle without human intervention.source：arXiv
Rippletide introduces decision context graphs to help enterprise AI agents remember
Rippletide, a startup in the Neo4j ecosystem, introduced decision context graphs to solve the enterprise AI agent forgetting problem. The framework gives agents structured memory, time-aware reasoning, and explicit decision logic. The key capability is non-regressivity: agents can freeze validated sequences of actions and compound on them over time. Unlike RAG which only surfaces relevant documents, decision context graphs encode applicability, temporal validity, and decision paths.source：VentureBeat
AI Investment & Startups
Hark AI startup raises $700M Series A at $6B valuation with chip giants backing
Hark, a one-year-old AI startup developing personalized AI devices, raised $700 million Series A led by Parkway Venture Capital with participation from Nvidia, Intel Capital, AMD, Qualcomm Ventures, ARK Investment Management, Salesforce Ventures, and Align Ventures. The San Jose-based company says it is building 'advanced personalized intelligence and next-generation hardware' and plans to release a product later this summer.source：Crunchbase News
MiRus medical device company raises $1.5B corporate round led by Boston Scientific
MiRus, a Marietta, Georgia-based company developing implants and treatment systems for musculoskeletal disorders, raised a $1.5 billion corporate round led by Boston Scientific. The deal comes with a 34% equity stake for Boston Scientific. MiRus has now raised $1.6 billion to date, as strategic investors continue betting on next-generation orthopedic and spinal technologies.source：Crunchbase News
Modal Labs raises $355M Series C at $4.65B valuation with $300M ARR
Modal Labs, a New York-based company providing serverless cloud computing tools and GPU access for running AI models and testing AI-generated code, raised $355 million in a Series C round led by General Catalyst and Redpoint with participation from Bain Capital Ventures, Menlo Ventures, and Accel. The round reportedly values the company at $4.65 billion. CEO Erik Bernhardsson told Reuters that Modal's ARR has soared to $300 million, up from about $60 million in September.source：Crunchbase News
Exa AI search startup raises $250M at $2.2B valuation led by Andreessen Horowitz
Exa, a San Francisco-based AI search startup developing AI-native search infrastructure for agents and large language model applications, secured $250 million at a $2.2 billion valuation in a Series C round led by Andreessen Horowitz. The latest raise brings Exa's total funding to $357 million, as competition intensifies around AI retrieval and search tools.source：Crunchbase News
Anthropic in talks with Microsoft to rent Azure servers with AI chips
Anthropic is in early talks to rent Azure servers with Microsoft's AI chips, according to The Information. The move comes as even the SpaceX $15 billion per year megadeal is reportedly not enough capacity for Claude. Anthropic has been steadily increasing its Azure usage. Microsoft's Maia 200 chips are designed to help run existing models like Claude, even if they are not as fast at helping to train new ones.source：The Verge
Digital Marketing Updates
Take It Down Act fully in force: platforms must remove deepfakes within 48 hours
The Take It Down Act's takedown provision is now fully in force as of May 19, 2026, requiring online platforms to remove nonconsensual intimate imagery, including AI-generated deepfakes, within 48 hours or face civil penalties of more than $53,000 per violation. The FTC sent letters to over a dozen companies including Amazon, Alphabet, Apple, Meta, Microsoft, TikTok, and X. However, free speech advocates and abuse opponents warn the law could lead to over-moderation and potential censorship.source：The Verge
Spotify launching AI-generated audiobooks for authors
Spotify announced it will soon offer authors a way to AI generate audio versions of their books. The feature will allow authors to convert their written content into audio format using AI voice synthesis technology, potentially democratizing audiobook production for independent authors who may not have resources for traditional narration.source：The Verge
ChatGPT PowerPoint integration launches for business and enterprise users
OpenAI and Microsoft launched a ChatGPT integration for Microsoft PowerPoint, adding a sidebar where users can create or edit presentations using chatbot prompts along with documents, images, and other source material. The feature is available now in beta for ChatGPT users with Business, Enterprise, Edu, Teacher, K-12, Free, Go, Pro, and Plus plans, following an earlier add-on for Excel and Google Sheets.source：The Verge
Aleksander Madry leaves OpenAI to focus on AI's economic impact
Aleksander Madry, one of OpenAI's top safety executives and former head of preparedness, announced he is leaving the company to work on something new centered on AI's impact on the economy. Madry was reassigned to a role focused on AI reasoning last summer before his departure this week.source：The Verge
Key Trends Today
AI Security Research AcceleratesAnthropic's Project Glasswing with 50+ partners found 10,000+ critical vulnerabilities in critical software, marking a 10x increase in bug discovery rates. AI is transitioning from creating vulnerabilities to helping fix them at unprecedented scale.
Enterprise AI Operations MaturesResolve AI and Rippletide tackle enterprise AI agent reliability with multi-agent debugging systems and structured memory frameworks. The focus shifts from building AI to making AI production-ready and non-regressive.
AI Infrastructure Funding Remains StrongThree major funding rounds exceed $300M this week: Hark ($700M), Modal Labs ($355M), and Exa ($250M). Infrastructure and inference speed (Cerebras 7x faster) remain key competitive differentiators.
Regulatory Landscape Takes ShapeTake It Down Act enforcement begins while Trump administration postpones AI executive order. Balancing AI safety, deepfake removal, and free speech remains a complex policy challenge.
Content sourced from public RSS feeds. AI-generated annotations and source attribution only. For commercial use or removal requests, please contact us.
Generated by eBaoGao: 2026-05-24 08:39:00. For more AI news, topics, and blogs, visit zanspark.com