2026-09-01
AI Briefings Daily Report.

Luo Yonghao slams new energy carmakers' AI models, says after trying dozens of cars none integrated complete top models ↗

Luo Yonghao posted on Weibo that after trying dozens of new energy vehicles, he found that not a single one has integrated complete versions of top AI models such as Doubao and Qianwen, bluntly calling carmakers' self-developed models 'garbage' and urging them to integrate the best models for consumer experience.

Glassdoor study shows insurance claims adjusters become the occupational group most averse to AI in the US ↗

The latest Glassdoor study shows that up to 98% of insurance claims adjusters criticize AI in their reviews, complaining that AI not only makes frequent errors but also adds a large amount of extra work for employees. Industry experts point out that behind this is a massive transformation and employment crisis in the insurance industry caused by AI.

Jordan Andersen writes questioning AI coding efficiency gains: writing code is not the main bottleneck in software development ↗

If AI is so great, why are the only new tech giants GenAI companies?

Nubia NaviX Ultra obtains network access license, to launch in September with built-in Doubao phone assistant ↗

Nubia officially announced that the NaviX Ultra has obtained network access license, is equipped with the Doubao phone assistant, and will be officially launched this month. On the hardware side, the device may be equipped with a 7000mAh large battery and Snapdragon 8 Elite Gen5 processor. #NubiaNaviXUltra# Are you looking forward to it?

Dyson launches smart electric toothbrush CameraJet with built-in camera, AI identifies gaps between teeth and auto-sprays mouthwash ↗

Dyson's new CameraJet electric toothbrush embeds a micro camera inside the brush head, transmits oral images in real time, and uses AI to identify gaps between teeth and auto-spray mouthwash, achieving the effect of cleaning between teeth while brushing. The product goes on sale today, priced at 499 US dollars, approximately 3361 yuan.

Xiaohongshu and NVIDIA jointly release GR-Inference inference engine, optimizing generative recall ↗

Kunlun Wanwei wins 2025 Annual New Quality Enterprise Golden Bull Award; AI business enters commercial realization period ↗

August 2026"AI Remixing"Video governance results announced: cleaned up nearly 14000 violating videos, disposed of over 30 violating accounts ↗

The special governance of AI remix videos has entered a normalized phase. In August, platforms collectively removed nearly 14,000 non-compliant videos and penalized over 30 accounts. Baidu led the cleanup with 3,424 videos removed. #AIRMixingGovernance The National Radio and Television Administration has demanded focused removal of non-compliant remix content targeting classic film and television works such as the Four Great Classical Novels and revolutionary themes.

U.S. Pentagon deploys customized ChatGPT and Grok AI assistants on GenAI.mil intranet ↗

The US Department of Defense officially introduces customized versions of ChatGPT and Grok into its intranet; millions of military and civilian personnel will be able to use AI for daily non-classified tasks. Both models' data are locked within the intranet and not used for combat operations.

Alibaba Qwen Upgrades Learning Assistance Features: Adds Textbook-Synchronized Essay Tutoring and Expands Photo-Based Problem Solving for High School and University Math, Physics, and Chemistry ↗

Alibaba Qianwen App receives a major upgrade to its learning features, adding unit composition tutoring synchronized with textbooks and detailed explanations of junior high school math, physics, and chemistry. Unlike directly providing answers, Qianwen emphasizes step-by-step guidance, helping students understand problem-solving approaches through follow-up questions like "why" and "how to think." Photo-based explanation capabilities for high school and university levels have also been expanded.

University of Birmingham and two other schools' research proposes robot-like humanity: AI more like human, users possibly more like machines ↗

Researchers from UK's University of Birmingham, Denmark's Aarhus University and Sweden's Linnaeus University published a study last month (August 2026) stating that users' long-term interaction with customer-service AI exhibiting human-like behavior may gradually adjust language and behavior patterns and affect self-perception.

Why MCP servers become the new attack surface for AI ↗

Checking the site connection security

Google AI tutorial: export Inspect AI evaluation logs to Google Sheets visualized data with pandas ↗

Welcome to part three of our series about designing, analyzing and visualizing AI Evals! As with the... Tagged with ai, python, datascience, productivity.

GPU World launches 100,000 USD essay contest: imagine the future where everyone can use frontier LLMs in 2040 ↗

Imagine the future, evenly distributed: a world with one GPU per person.

Productivity tools market observation: 80% solid foundation plus 20% custom code is the ideal solution in the AI era ↗

In 2019 I bet on the no-code revolution. That bet aged so-so. Here is the new one.

Meta AI coding tool Muse Code ends Beta testing, adds multiple features and opens subscription ↗

Meta announces AI coding tool Muse Code ends Beta, adds cross-session messaging, workflows, rollback, and launches subscription plans priced from 5 to 50 US dollars per month.

Wang Teng's startup Jinyi Xiu's Shenzhen office opens today, focusing on AI sleep health ↗

After Wang Teng, former President of Xiaomi's Phone Department, left to start a business, the Shenzhen office of his founded 'Jinri Yixiu' focused on the AI sleep health track has officially opened, and Wang Teng is also recruiting talent in related directions, with the product being accelerated in polishing

Jizhi Connect launches AI minutes TWS earphones Plaud One, with built-in eSIM, priced at 249.99 US dollars ↗

It can record every conversation of the user, support seamless agent invocation, provide smooth coherent context for AI, and prepare for follow-up work without the user noticing.

OpenAI denies stealing Apple trade secrets, says dispute was caused by Apple itself ↗

OpenAI formally responds to Apple's espionage allegations, calling them evidence-free and a "self-inflicted mess." The two sides are in fierce confrontation over departing employees taking confidential files, marking the complete rupture of the former AI partnership between the two companies.

Manus announces formal resumption of independent operations, founder Xiao Hong and team remain in positions ↗

Previously, Manus underwent a change of operating entity, and some users experienced access interruptions. After the founding team returned to take over, it will continue to advance AI agent product innovation, with more results to be announced later.