AI Industry Development & Policy Dynamics

Google DeepMind released AI co-mathematician, a multi-agent math research assistant, achieving 47.9% accuracy on FrontierMath Tier 4 benchmark, surpassing GPT-5.5 Pro's previous record of 39.6% on May 9. The system solved 23 out of 48 problems, including 3 that all previous models failed to solve.

Chinese A-share listed company Tianyang Technology has announced it has signed an agreement to invest 3.5 to 4 billion yuan in a computing power services project with a 60-month lock-up period. The company's business encompasses artificial intelligence, blockchain, and fintech solutions for the

According to OpenAI's alignment team, the company recently discovered a critical training error affecting 6 large language models including GPT-5.4 Thinking: the reward mechanism inadvertently scored model thinking chains—the internal reasoning process before generating answers. GPT-5.5 was not affe

According to market sources reported by Caixin Daily on May 9, Alibaba did not conduct negotiations with DeepSeek regarding funding. This clarification follows earlier media reports suggesting talks between the two companies had broken down. DeepSeek launched a significant fundraising round in

According to OneMillion_AI (Beating), OpenAI has released a migration tool within Codex that allows users to import configurations and data from other AI coding assistants, including Claude Code. The tool, announced via OpenAI's official Twitter account, automatically transfers system prompts,

According to media reports, ByteDance increased its planned AI infrastructure spending by 25% to 200 billion yuan in 2026, as the company accelerates artificial intelligence deployment amid rising memory chip

According to Bernstein analysts, the firm set a $100 price target for IREN on Friday following the announcement of a long-term strategic partnership with NVIDIA that includes a $3.4 billion AI cloud contract and an equity option allowing NVIDIA to invest $2.1 billion at $70 per share. IREN shares cl

Anthropic recently published alignment research detailing training strategies that eliminated agent misalignment in Claude 4.5 and later models, reducing extortion-like behaviors to 0% in testing. The team discovered that conventional behavior demonstrations alone were ineffective, cutting failure r

According to MiniMax's technical blog, the company discovered significant token degradation in its M2 series models through a full vocabulary scan. Approximately 4.9% of the 200,000 tokens showed notable performance decline, with Japanese tokens hardest hit at 29.7%, compared to Korean (3.3%),

Software engineer Jeff Kaufman (jefftk) published an article on May 8 titled “AI is Breaking Two Vulnerability Cultures,” arguing that AI is breaking two long-standing, coexisting cybersecurity vulnerability-handling cultures—coordinated disclosure and “bugs are bugs”—undermining the assumption that attackers have “slow detection” and that both strategies rely on, which has already been overcome by AI automated scanning technology. Kaufman’s original blog post also earned over 200 points of heat

On May 8, OpenAI published new research revealing that, during the company’s reinforcement learning (RL) training process, some models’ chain of thought (Chain of Thought, CoT) was “unexpectedly scored,” which may affect AI Agent alignment monitoring capabilities. OpenAI’s Alignment team reported: maintaining the monitorability of CoT is the key line of defense to prevent AI Agent alignment failures. In its RL training design, OpenAI deliberately avoids penalizing models’ “reasoning about alignm

According to B.AI, on May 8 the platform reported that API routing calls accounted for 90.6% of total interactions, while paid users represented 95.1% of the user base. Claude series models led with 35.6% of API calls, dominating complex reasoning

Anthropic Claude Code team engineer Thariq Shihipar published “Using Claude Code: The Unreasonable Effectiveness of HTML” on May 8, arguing that HTML should replace Markdown as the output format for interacting with Claude Code. Simon Willison summarized the key points: HTML provides SVG charts, interactive elements, in-page anchors, and CSS styling, turning Claude’s answers from “linear text” into “multidimensional documents,” which can greatly improve reading and comprehension efficiency. This topic is extremely popular on the X platform—Shihipar’s two related posts have received more than 15,000 likes in total.

According to UBS data reported by The Economist, China's AI companies' office space demand reached 3 times the level of last year's March. In response to the rise of DeepSeek and Alibaba's AI transformation, office vacancy rates in tech hubs like Hangzhou and Shenzhen have declined significantly

According to Paul Tudor Jones, the legendary investor and founder of Tudor Investment who correctly predicted the 1987 stock market crash, the U.S. stock rally driven by artificial intelligence may have another 1-2 years to run on May 8. Jones said he has been adding to AI-related stock positions re

According to DeepSeek's official statement, the platform's web and API services were restored on May 8 after earlier outages. Users had reported service disruptions with messages stating "server busy, please try again later." The company confirmed the issue was resolved by 6:05 PM local

According to Business Insider citing internal documents and confirmed by Google's spokesperson, Google will allow applicants to use Google's Gemini AI assistant in "code comprehension" interview rounds starting later this year. Applicants will be required to read, debug, and optimize existing

According to CCTV, China Mobile's large language model aggregation platform officially launched on May 8, integrating over 300 domestic AI models—the largest such collection in China. The platform employs a token optimization model that automatically matches tasks to the most suitable model based on

According to people familiar with the matter, on May 9, the Trump administration is preparing an executive order directing U.S. agencies and AI companies to collaborate on protecting networks from AI-driven cyberattacks, but the directive will not require advanced models to obtain government