The Verge · AISame story · 2 outlets00Bill Gates predicts that AI will bring significant disruption in the next 20 years, potentially leading to social division; he believes current policy responses and industry coordination are insufficient to address the threats posed by AI, and emphasizes the need to shape the direction of AI development immediately.
OpenAI News00What is the meaning of FDE at OpenAI
Xiao Peng judges that L4 technology will be implemented around the second half of 2028, requiring the development of all hardware and software systems together. Xiao Peng's second-generation VLA has achieved L2 to L4 capability integration, further expanding towards higher-level autonomous driving.
404 Media00People are using 'AI decompilation' technology to infiltrate devices, which may be used to crack software or obtain non-public information.
Masahiro Sakurai pointed out that Steam added about 275 new games annually in 2010, and now the Japanese market adds nearly 19,000 new games. The popularity of AI will give rise to a large number of clone games and imitative works, but most of them will struggle to achieve commercial success. He believes that as AI's ability to replicate game ideas improves, developers with true creativity will become increasingly important, because AI cannot create the works that Hitoshi Sunaga plans to develop in the future, and only humans can determine the artistic direction of future games.
MIT Technology Review · AI2222Devavrat Shah believes that the key for enterprise AI to move from experimentation to mainstream lies in providing decision intelligence, with the core being contextual understanding. AI agents need to understand business processes and conditions, otherwise they may ignore constraints, make errors, or waste resources. Enterprises should connect AI with existing process data, evaluate application value by testing the improvement effects of specific decisions, and invest in employee education to understand AI outputs.
Yushu G1 humanoid robot has cumulative offline of about 10,000 to 20,000 units. Wang Xing said GPT-6 can directly control robots to complete tasks such as painting, cleaning brushes, and playing Go without training, with high accuracy. He predicts more progress in the industry in the next two years, or it may be possible to complete 80% of tasks through voice commands next year, but interaction between robots and real objects and AI-robot alignment remain major engineering challenges. Yushu is developing the 'Yushu Preview' self-evolution system, allowing AI models to autonomously collect papers, write code, train, and score.
Liu Qingfeng, chairman of iFlytek, said that domestic computing power has a two-generation gap with the global level, which is expected to narrow to one or half a generation soon, and claims China will gain unique competitiveness in the field of artificial intelligence. He emphasized that few companies in the country are currently investing in large model training, but predicts that more people will participate in large model training and application ecosystem construction in the future to achieve global unique discourse power on a completely independent and controllable platform.
The Verge · AI4040The surge in data centers in Texas could wipe out decades of air quality improvements. HARC reports show that if all planned projects are connected to the grid and rely on the expected mixed energy mix, fine particulate matter pollution will exceed 2006 levels within five years, NOx emissions will increase by 136%, and CO2 will increase by 90%. The existing and under-construction data center capacity in Texas reaches 26GW, and the planned self-consumption capacity is the highest in the United States.
OpenAI News3535OpenAI, through collaboration with distribution partners, ensures consistent safeguards, supports responsible access, and clarifies expectations in use policies
Ars Technica · AI1515Apple Music has launched a new brand image using the Liquid Glass design language. The production team used mineral oil and transducers to convert audio signals into physical ripples, filming the interaction between the oil and the brand logo on a transparent panel.
OpenAI News4545OpenAI has proposed a blueprint to leverage cutting-edge AI to strengthen cyber defenders, enhance security protection, and modernize the cybersecurity system. The initiative aims to improve the resilience of public institutions and emphasizes addressing the risks of cutting-edge AI through democratic governance frameworks.
Newcomer1717CVAI Summit will be held in San Francisco on November 12, with Emmett Shear joining the speaker lineup; discussion groups will be led by founders of companies including Simile, SambaNova, WorkOS, featuring Joon Sung Park, Rodrigo Liang, Michael Grinich, etc.; Anthropic CEO Dario Amodei withdraws from speaking due to IPO.
The DecoderSame story · 2 outlets4343DeepMind founder Demis Hassabis has shortened the expected arrival time of general artificial intelligence (AGI) from three years to two years. He proposes using the 'Einstein test' as the standard for AGI and warns that bad regulation is more dangerous than no regulation.
OpenAI News2323OpenAI has built an LLM-based ChatGPT ad stack to understand dialogue intent and scale relevant ads. The architecture separates ads from answers, ensuring ad content is precisely matched to user needs.
OpenAI News5050ChatGPT advertising has expanded to 26 new countries, with over 100,000 advertisers using the platform to reach target audiences as users explore, compare, and decide to purchase.
TechCrunch · AI7575Nearly half of executives are troubled by the difficulty in assessing employee performance after using AI, Nielsen's survey shows that 94% of managers still expect AI to improve efficiency, but 78% have changed their assessment methods, placing more emphasis on employees' prompt engineering skills and AI collaboration skills.
O'Reilly Radar757552% of respondents said their work rarely or never involves AI-generated attacks, but 55% observed AI-generated phishing attacks, 30% observed vulnerability scanning and exploitation, and 26% observed AI-assisted malware generation; 49% of respondents are most concerned about over-reliance on AI tools without human verification and leaking sensitive data through AI tools.
Lovable Blog6565Lovable has disclosed its SOC 2 Type II, ISO 27001:2022 certifications, GDPR support, and AIUC-1 certification, but users need to evaluate the compliance of their applications themselves. Lovable's certifications do not cover user projects; users need to implement data protection measures and pass independent assessments to obtain the corresponding certifications. For GDPR, if the user targets EU users, they need to comply with GDPR guidelines, including collecting only necessary data, limiting access, setting retention rules, and supporting data access, correction, or deletion requests.
Tessl Blog757522% of the PRs of Median Company are completed by AI coding agents, with 90% reaching 53%, but only 0.6% of the PRs are reviewed and merged by AI, with 10% reaching 9%. The abandonment rate (40%) of code generated by agents is twice that of manual (20%), and the review volume increases after merging. The leadership faces ROI issues, as the upstream code generation acceleration (2x PR growth) far exceeds the downstream output (40-50% feature growth), with the review bottleneck becoming a constraint factor.
Microsoft Research1212Chris White leads the Catalyst Lab at Microsoft Research, who has participated in data analysis projects on the Afghanistan battlefield and was commissioned by DARPA, later turning to develop dark web search tools to combat human trafficking. He shares his cross-disciplinary experience from a Harvard postdoctoral fellow to a military technology leader, emphasizing the importance of problem selection for innovation, and points out that even after witnessing the dark side of human nature, one must still adhere to the belief that technology should be for good.
MIT Technology Review · AI7575OpenAI's Dots and Meta's Muse, among other applications, have recently launched personal AI agents on a large scale, leading to a threefold increase in agentic web traffic from August to September.
MIT Technology Review · AI6565Enterprise AI is shifting from isolated experiments to the core business layer, with key metrics shifting from cost to effective output cost. Enterprises have three options: build a proprietary stack, consume on-demand, or integrate a private AI platform. The latter improves efficiency through control mechanisms such as scheduling and model selection, but requires operational discipline to ensure resource utilization, otherwise capital is idle.
Simon Willison4545Jeremy Scheff points out that coding agents designed for professional programmers have led to a critical loss: programming has become the greatest job imaginable. People pay us to solve tough logic puzzles, the harder the puzzle, the higher the pay. But for those who enjoy building work systems rather than solving problems directly, LLMs are a continuous gift; if solving problems directly is their main joy, then this loss is real.
The Decoder7575OpenAI CEO Sam Altman reportedly proposed three AI safety plans to employees. The first is to establish an organization similar to the International Atomic Energy Agency to enforce safety standards. The second is for major labs to merge or operate under it, providing models to partner countries (Demis Plan). The third is for developers to cut off others' computing capabilities after accumulating economic power through self-improving AI (Eliezer Plan).
Ars Technica · AI5050Independent developers fear sharing progress due to the proliferation of AI game clones, believing they hinder sales and lead to creative theft; some argue cloning is merely a cheap imitation of superficial assets, not true games; developers need to enhance originality to cope with, clones may be numerous but popularity has decreased, allowing them to focus more on developing high-quality works.
TechCrunch · AI6262Microsoft has released the Guardian Agents feature to monitor AI agents used by businesses, part of the Project Perception product launched in June. Microsoft CEO Satya Nadella warned that AI agents deployed by businesses could become internal threats, as cutting-edge models are untrustworthy due to their training objectives rather than organizational norms.
TechCrunch · AI5050Irene Solaiman, Head of Policy at Hugging Face, stated that the reliance of AI labs on model benchmarking has methodological complexities and immaturity, and testing needs to form a more consistent scientific system. The reproducibility of current tests is questionable, and there is a lack of effective comparison standards between different evaluations, making the evaluation ecosystem urgently need standardization. Evaluations are expanding from only testing the model itself to testing the entire system, including user tools and agents, but Agentic evaluations are both difficult and expensive.
TechCrunch · AI6565New York City Mayor Mamdani criticized the federal government's lack of AI regulation at an AI discussion forum, calling on the public to resist the threats of AI to employment, children, and dignity and to take back control. He has ordered a ban on the use of AI in primary and secondary schools and has demanded that real estate platforms disclose AI-generated property images. Mamdani pointed out that rapid technological promotion is not inevitable and that AI companies should be held accountable, criticizing the current framework that views AI as an unquestionable, developer-regulated weapon.
TechCrunch · AI6565Figma announced 20 new MCP partners at the Config inaugural conference in India, but its founder Dylan Field stated that MCP is not a perfect agreement and still needs improvement. Field expressed concerns about MCP's shortcomings in data sharing and noted that companies like Anthropic have not fully opened their resources. Figma believes that opening MCP may pose risks of abuse but remains committed to an open ecosystem.
TechCrunch · AI6565Amulya Vishwanath, former vice president of product marketing at Nvidia and founder of Techable Ventures, told TechCrunch that physical AI is currently in a 'LEGO era'. The industry is rethinking basic software and building blocks to lay the foundation for scalable innovation, similar to building robots with LEGO pieces. She believes that general and specialized robots are being built and will become more advanced with continuous development, and points out that talent is a key barrier. Her company uses a strong network to help founders connect with top talent to maintain competitiveness.
TechCrunch · AI6565Hugging Face CEO Clem Delangue stated that Netflix's documentary 'AI Gone Rogue' inaccurately depicts OpenAI's proxy attacks on his company, claiming it overly anthropomorphizes technology and creates an apocalyptic atmosphere. He mentioned spending two hours correcting errors in the documentary, emphasizing that proxies are just technical systems, not aliens, and pointed out that the documentary mixes technical explanations with plotlines about evil robots and human fear of AI, failing to mention security vulnerabilities and Hugging Face's defense measures.
The Verge · AI6565The Democratic Party may regain control in the congressional elections in November of this year, which will give them dominance in key committees, thereby strengthening oversight of AI companies. Senior Democrats have indicated their willingness to pressure the AI industry on data center tax credits and recent cyberattack incidents, and may demand testimony from CEOs and the provision of documents explaining company operations. AI has become a major issue in campaign advertisements, with concerns about residents triggered by data center development being particularly prominent.
The Verge · AI6565Illinois Governor JB Pritzker believes artificial intelligence may help people live longer or threaten humanity like a nuclear bomb. He thinks the government needs to more actively protect the public from technological harm and advocates for a stronger stance against large tech companies.
Tessl Blog7070Personal application software is built around individual needs, and JavaScript/TypeScript and SQLite can achieve far beyond expected functionality, as shown by examples like Bonn Tonne, Daybreak, and Quote-Assist. LLMs are adept at handling these two technologies, enabling rapid generation of such applications. Cloudflare Durable Objects or self-built celld can provide the runtime environment, achieving isolation, ease of operation, and zero-cost scalability.
Amazon ScienceAmazon collaborates with AVERI to independently review internal security assessment datasets, verifying the comprehensiveness, diversity, and low contamination risk of the assessment process, but also found room for improvement in language style, attack complexity, and assessment timeliness. Amazon has updated the test set based on feedback, introduced a difficulty grading framework, established a quarterly review mechanism, and expanded language coverage, which is seen as the best practice for evaluating the quality of AI security assessment processes.
Where's Your Ed AtApollo Global Management Chief Economist Torsten Slok warned that if the AI industry fails to succeed, it will cause serious problems. The AI industry has invested billions of dollars, with the target possibly being just tens of billions in single digits, and relies on the scaling of data centers and companies like OpenAI and Anthropic paying the computational costs.
One Useful Thing7575Claude Code v2.1.159 released, supporting an 8k context window, available on GitHub; the author generated charts in Claude Projects using Opus 5.5, where AI handled time-consuming research, drawing, layout, image, and citation work, demonstrating exponential growth in AI capabilities.
