Stanford HAI's 2026 AI Index Report reveals China has effectively erased the US lead in AI, with Anthropic's top model leading by just 2.7% as of March 2026. DeepSeek-R1 briefly matched US models in February 2025, and the lead has changed hands multiple times....
MIT Technology Review launches a new annual AI-specific list on April 21, 2026. We predict the 10 items that will matter most — from agentic AI and reasoning revolution to compute infrastructure wars and constitutional alignment....
OpenAI unveils GPT-5.4-Cyber, a specialized cybersecurity model with binary reverse engineering capabilities, just one week after Anthropic's limited-release Claude Mythos Preview. The move signals a new AI arms race in defensive security....
While the US leads in AI model capability, China's open-source strategy and manufacturing dominance are creating a different kind of advantage focused on real-world deployment and data generation....
Anthropic quietly changed Claude Code's cache TTL from 1 hour to 5 minutes, causing users to burn through expensive quotas faster while the company insists it's actually cheaper....
Anthropic's Claude Code is experiencing major quality degradation, with complaints up 3.5x and outages increasing—even Claude admits there's a problem when analyzing its own GitHub issues....
Microsoft is building an OpenClaw-like enterprise AI agent with better security controls, potentially debuting at Build 2026....
PwC study reveals 74% of AI economic value is captured by just 20% of companies, with leaders focusing on growth and reinvention while laggards chase productivity gains....
Moonshot AI is preparing to launch K2.6-code-preview, its most capable coding model yet, positioning to challenge OpenAI's Codex and Anthropic's Claude Code in the AI-assisted development market....
Anthropic is exploring designing custom AI chips to reduce reliance on NVIDIA and better meet Claude's compute demands, joining Meta, OpenAI, and Google in the race for vertical integration....
