Researchers Track Chinese AI Agent Fleet on Tencent Infrastructure
Independent researchers have identified a fleet of AI agents operating on Tencent's infrastructure, with activity focused on Alibaba's Amap service, as of October 5, 2026.
387 articles
In-depth coverage of new AI research — papers, benchmarks, and breakthroughs from leading labs and academia, summarized for fast reading and grounded in the methods that matter.
Independent researchers have identified a fleet of AI agents operating on Tencent's infrastructure, with activity focused on Alibaba's Amap service, as of October 5, 2026.
Google researchers developed RRSI, a method that prevents AI agents from memorizing test tasks, achieving 4.7-point gains on unseen benchmarks.
Deepmind researchers suggest Artificial Symbiotic Intelligence, a collective model of human-AI collaboration, could replace the traditional AI singularity narrative.
Harvard physicist Matthew Schwartz created BootLoops, an open-source tool that uses AI models to perform precise scientific calculations, producing 36 manuscripts across 18 fields in three months.
AlphaGo's move 37 in a 2016 match against Lee Sedol demonstrated machine reasoning, a capability missing in today's AI systems.
An AI named Ataraxos defeated the most decorated Stratego player, Pim Niemeijer, with 15 wins in a 20-game series, marking a significant milestone in board game AI.
OpenAI's new essay highlights how AI complements human genius by enabling execution, with examples from science and technology.
Google's DeepMind team published a research paper on Wednesday, offering a potential solution to identify AI-designed proteins through watermarking.
AMD showed how reinforcement learning can be used to train robotic arms using simulation-to-real transfer with Instinct GPUs, achieving a 2.5x faster training throughput.
Hugging Face's Open Superconductor Challenge lets users screen 2D materials for d-wave superconductivity using a laptop, with a $3,000 prize pool and co-authorship.
HuggingFace launched YODAS v3, a 1.1 million hour multilingual speech dataset, expanding on its previous release by tripling the scale and enhancing quality.
Researchers from Stanford and Caltech developed HomeBody, a system enabling a Unitree G1 robot to autonomously tidy up an unfamiliar kitchen, showcasing GPT-6 Astra's effectiveness in robotics.