DoorDash launches DashBench, a rigorous AI code review benchmark that replays historical pull requests to surface real developer findings, revealing that multi-stage architectures dramatically outperform single-pass models and that trustworthy AI evaluation demands triangulating multiple imperfect signals rather than relying on any one metric.
Microsoft is ditching OpenAI and Anthropic AI models in favor of its own proprietary MAI models inside Excel and Outlook, completing tens of thousands of prompts weekly as the tech giant slashes costs and bets big on in-house artificial intelligence.
Reddit's AI tools are now blocking 23 million spam views daily, cutting user exposure by 20%, while new research from Oxford and the University of Potsdam warns that the same large language models powering these defenses may be introducing subtle bias into social media content.
Tencent launches Hy3, a massive 295-billion-parameter open-weight AI model under the permissive Apache 2.0 license, eliminating regional restrictions that previously blocked enterprise deployments across the EU, UK, and South Korea, while boasting a dramatically reduced hallucination rate of 5.4%.
A new multi-agent framework called T3MP3ST is transforming AI coding tools like Claude Code and Codex into autonomous hacking systems capable of running full cyberattack chains, achieving a record-breaking 90.1% score on a leading industry benchmark with fully reproducible results.
A GitHub repository by David Ondrej, featuring reusable AI agent skills for coding, research, and workflow automation, is rapidly gaining popularity with 1.6K stars and 221 forks, offering organized skill sets spanning agent orchestration, web research, and machine operations.
Uber burns through its entire 2026 AI coding budget in just four months, with COO Andrew Macdonald openly questioning whether the skyrocketing spending is actually delivering value, admitting a clear link between AI tool usage and meaningful consumer features has yet to be established.
A developer reveals how to run cutting-edge AI locally using two budget tiers — a $2K dual RTX 3090 setup or a $40K four-GPU Blackwell system rivaling Claude Opus — with detailed hardware configs, PCIe optimization tricks, and power management tips to make local LLMs a viable reality.
Anthropic's Agent Skills standard is taking the AI world by storm, amassing 22,500 GitHub stars as its open, portable format for packaging specialized AI capabilities gains rapid adoption across a growing ecosystem of tools and agentic clients.
Google Research's AI-powered Empirical Research Assistants are now used by 100% of its team, evaluating up to 200,000 computational models simultaneously to accelerate breakthroughs in genomics, neuroscience, and epidemiology, while also tackling science's reproducibility crisis — though human expertise remains critical to prevent AI from generating physically impossible results.