AI daily

2026-09-11

2026-09-11 AI Daily

Featured

APriority77
The AI Insider
1 sources

Graph AI Raises $13.3M Series A to Scale AI-Native Patient Safety OS

Graph AI announced a $13.3 million Series A round led by Insight Partners, with participation from existing investor Bessemer Venture Partners. The company develops Graph Safety, an AI-native patient safety platform that combines AI with deterministic controls, validation layers, and audit trails for regulated pharmacovigilance environments. Graph AI says the platform reduces case processing turnaround time from over three hours to under 10 minutes and lowers operating costs by up to 66% in live deployments.

APriority77
AI前线
1 sources

GOAI World AI Open Source Competition Announces Top 70 Finalists

The Global Open-source AI Competition (GOAI) has announced its top 70 finalists across four tracks: New Intelligence Infrastructure, Boundless Applications, Frontier Exploration, and Embodied Future. A total of 2,999 initial entries were received, and the 70 selected projects will compete in the finals and GOAI DAY on September 22-23 in Hangzhou for track championships and the grand prize.

APriority77
阿里云开发者
1 sources

Alibaba Cloud Token Plan Personal Edition Upgrade: More Features at the Same Price, Adds 12 Agent Harness Tools

Alibaba Cloud upgraded its Token Plan Personal Edition on September 11, adding 12 categories of Agent Harness tool benefits to the Standard and Pro plans without changing prices or Credit quotas. The tools cover search, web parsing, image generation, speech processing, and code execution, and are exposed via the MCP protocol. The plan also adds five postpaid services—video generation, Managed Agent, RAG knowledge base, Memory, and Sandbox—expanding the subscription from model credits to a combined model-plus-Agent service offering.

APriority81
机器之心
1 sources

iFlytek Releases Spark X2.5: Trained on 10,000 Domestic GPUs, AI Feeds Back into Compute Optimization

On September 7, iFlytek officially released the Spark X2.5 large model, featuring a MoE architecture with 293B-A30B parameters, enhanced code and agent capabilities, and support for over 200 languages. The model was pre-trained and post-trained on a 10,000-GPU domestic 910B cluster, with online inference on domestic platforms and multi-step speculative decoding. Notably, Spark X2.5 has begun participating in NPU operator optimization, allowing AI to feed back into underlying compute optimization and providing practical insights for engineering challenges in domestic large-scale training, such as operator efficiency, long context, communication, and stability.

APriority80
机器之心
1 sources

Beihang, NTU team proposes PhyFilter in npj Robotics to improve robot generalization beyond data scaling

A team from Beihang University, Beijing Aerospace Control Instrument Research Institute, and Nanyang Technological University's MARS group published PhyFilter in Nature npj Robotics. The module treats a neural network's learning error as a low-frequency signal that can be filtered out, using the robot's real-time state feedback and known physical differential structure to correct learned outputs online; it is plug-and-play, model-agnostic, and its filter parameters can be learned automatically. Validated on quadruped robots, drones, aerial manipulators, and acceleration sensing, policies trained only in flat-ground simulation transferred directly to stone, grass, sand, and gravel, and an aerial manipulator achieved centimeter-level grasping under 5 m/s wind disturbance, suggesting physical filtering feedback as an alternative generalization path to data scaling.

APriority80
InfoQ
2 sources

Wang Mengdi at Bund Summit: AI Has Not Yet Discovered New Basic Science

On September 10, Wang Mengdi, director of Princeton University's Center for AI Innovation, delivered a keynote at the 2026 Inclusion·Bund Summit main forum, arguing that while large models have absorbed vast human knowledge, they remain clearly limited in making original discoveries in basic science because they excel at finding the most likely answer in a probability distribution, whereas true discoveries often lie in the long tail. In collaboration with Princeton sociology professor Xie Yu, her team used large models such as ChatGPT and Claude for "simulated life" experiments, finding that models overestimate the most common outcomes and underestimate rare ones. Wang argues that the key to AI-driven autonomous scientific discovery lies in making the real world more verifiable, and her team is building an automated experiment platform and LabOS, an intelligent operating system for research laboratories.

APriority80
量子位
1 sources

Ant Bank Launches Bailing 2.0: World's First Inclusive Finance Agent for 42 Million Small Businesses

Ant Bank unveiled its AI banking progress at the 2026 Bund Conference, launching Bailing 2.0, the world's first inclusive finance agent for small and micro businesses, now available to 42 million merchants. The agent can understand vague requests through multi-turn dialogue and provide personalized credit solutions based on business data, such as raising a merchant's credit limit from 150,000 to 400,000 yuan in just 10 minutes. Behind the scenes, Ant Bank has also deployed AI workbenches like Qianliyan and Dinghaizhen to integrate AI into core risk control and approval processes.

More

APriority76
量子位
1 sources

Meta Loses Another Star AI Researcher Despite $1.5B Pay Package

Andrew Tulloch, a core researcher at Meta's TBD Lab, has left the company just two days after the launch of Muse and is set to join Anthropic's inference and performance team to optimize Claude Code training and improve model responsiveness. Tulloch had returned to Meta last year on a compensation package worth about $1.5 billion and worked on Muse-related R&D. His departure follows that of another high-profile AI hire, Yu Jiahui, highlighting Meta's difficulty retaining top AI talent despite lavish pay offers.

APriority75
InfoQ
1 sources

Kotlin 'God J' Two-Hour Controversial Interview: I'd Rather Lose 80% of Job Opportunities Than Use AI Programming

Jake Wharton, a prominent Android and Kotlin open-source developer, has stated that he refuses to join AI companies, work on AI-centric products, or accept mandatory AI usage, estimating this criterion excludes about 80% of potential employers. In a recent two-hour interview, he also criticized cryptocurrency and the forced adoption of large language models in software development, arguing that AI should only be a tool, not the foundation of engineering capability.

Selected
10
Sources
7
Featured
8
More
2
Previous2026-09-10Next2026-09-12
All reports