AI Briefing
TLDR AI news
The latest AI and developer news from TLDR AI, with short summaries.
Visit website →OpenAI Updates GPT-5.5 Instant for a More Natural and Useful ChatGPT
OpenAI has released GPT-5.5 Instant with improved ability to grasp question intent and handle complex constraints.
Anthropic Accuses Alibaba of Trying to Steal Its Technology Through AI Model Distillation
Anthropic has accused Alibaba to the US Congress of carrying out a large-scale distillation attack that mobilized thousands of accounts to steal the performance of its AI models.
Orca (GitHub Repository)
Orca, an AI orchestrator that lets you run and manage multiple AI agents in parallel, has been released.
Extracting Triangle Splats from Video Diffusion Latents
FLAT directly maps video diffusion model latents to triangle splats, producing accurate 3D scene geometry.
Notes on the Amazon vs. Perplexity Lawsuit
Amazon has filed a lawsuit against Perplexity, claiming its AI agent 'Comet' conceals user information and creates security risks.
GLM-5.2 Marks a Leap Forward for Open Agents
GLM-5.2, released by Z.ai, is an open-weight model that nonetheless delivers agentic performance competitive with top-tier models.
Gemini 3.5 Flash Introduces Computer Use Capability
Computer Use capability is now built into Gemini 3.5 Flash, enabling the construction of agents that perform tasks directly in browser and desktop environments.
Gemini researchers move to Anthropic
Key researchers who led the development of Google's Gemini models are successively moving to Anthropic and OpenAI, intensifying the talent drain.
Sakana Fugu Launches Scoring 93.2 on LiveCodeBench After Blocking Claude
Sakana Fugu launched scoring 93.2 on LiveCodeBench, outperforming Fable.
Tencent Testing AI Assistant in China's Largest App to Catch Up with Rivals
Tencent has introduced the AI assistant 'Xiaowei' within WeChat, moving to expand its AI services by leveraging its user base.
Anthropic preparing Cowork support in mobile app
Anthropic is preparing mobile support for Cowork, its agentic system for knowledge work, along with a model selection feature for Voice mode.
claude-sonnet-5
A slug named 'claude-sonnet-5', presumed to be a next-generation model, has been discovered in one of Anthropic's partner services.
Anthropic May Require ID Verification from Claude Users
Anthropic announced that it may require Claude users to submit ID documents to prevent account misuse and ensure regulatory compliance.
Moebius: A 0.2B Lightweight Image Inpainting Framework Achieving 10B-Class Performance
Moebius is a lightweight inpainting framework that, through LλMI blocks and an adaptive distillation strategy, reduces parameter count to under 2% while delivering performance comparable to large-scale models.
Model Size Scaling Outlook 2023-2031
Based on HBM bandwidth and pipeline architecture, this forecasts how model parameter sizes will change through 2031.
GLM-5.2 Raises the Bar for Open Models
GLM-5.2 comes close to frontier models in benchmark performance, pushing the boundaries of open models further.
Knowledge Agents: How to Outperform Frontier Models with Better Structure
Knowledge Agents injected with specialized knowledge outperform general-purpose frontier models in specific domains.
Claude Code's 'Extended Thinking' output is not the actual reasoning content
Claude Code's reasoning logs are merely encrypted signatures or summaries, not a complete record of the model's actual reasoning process.
Alibaba's AI Video Model Leaps to Global No. 2, Overtaking OpenAI Sora and ByteDance Seedance
Alibaba's HappyHorse 1.1 has emerged as a new powerhouse in the AI video market, ranking near the top in benchmarks.
AI Engineer Claims to Have Deciphered Linear A
Self-taught AI engineer Tom Di Mino claims to have deciphered Linear A, the Minoan civilization's script that had remained an unsolved problem for over 100 years.