AI Briefing
TLDR AI news
The latest AI and developer news from TLDR AI, with short summaries.
Visit website →Anthropic Revamps Enterprise Data Retention Policy
Anthropic has changed its policy to allow enterprise customers to retain data for 30 days on their own cloud infrastructure.
Vercel Launches Hacker Challenge Offering $1 Million for Sandbox Escapes
Vercel is running a two-week public hacking challenge that pays up to $1 million to researchers who discover sandbox isolation bypass vulnerabilities.
Warp's New System: Off-the-Shelf Software Factory for AI Development
Warp launched 'Warp Factories', an infrastructure system that simplifies building and operating AI software factories.
Anthropic Prepares to Grant Founders Super-Voting Rights Ahead of IPO
Anthropic plans to grant founders special voting rights to secure independence from shareholder pressure ahead of its IPO.
Introducing Harvey II
Harvey has launched Harvey II, an AI platform that automates the entire lifecycle of legal work.
From Zero to One (1-minute read)
Etched raised $700 million at a $21 billion valuation and delivered its first rack to Jane Street.
A bee does not fly like an airplane. Neither does AI.
Qwen3.6-35B is faster in speed but has longer inference times, resulting in slower final response times.
Nvidia's AI Moat Shifts from Chips to Capital
Nvidia is strengthening its dominance in the AI infrastructure market by leveraging its financial power alongside its chip technology advantage.
Reassessing the Data Moat
Reexamines the premise that algorithmic progress, rather than data volume, is the primary driver of AI model performance improvements.
Policy Algebra for Trust-Preserving Agentic AI Execution
This study proposes a policy algebra framework for trust-preserving agentic AI execution.
Fool's Gold (18-minute read)
Fool's Gold is a defense technique that trains the model itself to output fake information even when an attack succeeds.
Miles v0.1: Production-Level Post-Training (20-minute read)
RadixArk released Miles v0.1, a full-stack system for frontier-scale RL training.
FreeToken: Efficient Edge-Native MoE Serving via Bandwidth-Adaptive Execution
FreeToken provides an edge-native serving system that efficiently runs large MoE models by leveraging the heterogeneous resources of personal devices.
New US-made AI Model Designed for Customization
Thinking Machines released its first self-developed model, Inkling, under the Apache 2.0 license.
Building a Production-Grade Agent Loop
Liquid AI conducted an experiment to verify the autonomous production problem-solving capabilities of coding agents.
Git at Every Scale
Analyzes large-scale hosting strategies to overcome the structural limitations of distributed Git.
GLM-5.3 API Launch: $1.40/$4.40 per 1M Tokens (2-min read)
z.ai's GLM-5.3 has launched on the API at the same pricing as before, improving developer accessibility.
Qwen3.8-27B-Uncensored-MLX (4 min read)
Released an MLX quantized version of Qwen3.8-27B with safety alignment removed for Apple Silicon.
OpenAI GPT-5.6 Sol: 50% Discount on OpenRouter
A 50% discount applies to GPT-5.6 Sol API usage on OpenRouter.
Teaching Everyone Token Fishing
An analysis of Nvidia's strategy to maximize inference chip demand by building an open-source model ecosystem.