A 23-year-old amateur with no advanced math training cracks a 60-year-old mathematical conjecture using a single ChatGPT prompt, producing a novel proof that stumped world-class mathematicians for decades and is now being refined by leading experts like Terence Tao.
Anthropic launches a beta memory feature for Claude Managed Agents, allowing AI to retain and share context across sessions, with early enterprise adopters like Rakuten already reporting a 97% reduction in first-pass errors.
OpenAI launches GPT-5.5 just seven weeks after GPT-5.4, achieving a recursive self-improvement milestone where the model helped build itself, while introducing enhanced agentic capabilities across 44 occupations and expanded safety protocols with zero latency impact.
The Trump administration accuses China-backed actors of running an 'industrial-scale' AI theft campaign, using proxy accounts to steal capabilities from American models like Claude and Gemini, with OpenAI and Anthropic confirming attacks by firms including DeepSeek ahead of Trump's upcoming Beijing visit.
OpenAI launches GPT-5.5, its most advanced model yet, claiming it outperforms rivals from Google and Anthropic while marking a major step toward a unified AI 'super app' combining ChatGPT, Codex, and an AI browser for enterprise users.
MCP (Model Context Protocol) surpasses 300 million monthly SDK downloads, cementing its role as the dominant integration standard for production AI agents while powering key Claude features and offering developers best practices to slash token usage by up to 85%.
Qwen's new 27B model stuns the AI world by outperforming a massive 397B model across all major coding benchmarks while running locally on consumer hardware at just 16.8GB, generating complex code at 25 tokens per second.
Apple and Google are uniting on AI as Google Cloud powers Apple's next-generation foundation models built on Gemini technology, with a more personalized Siri launching later this year that taps into users' messages and emails through a redesigned Dynamic Island interface.
Unauthorized users breach Anthropic's powerful new Mythos AI model — a system the company warns could enable dangerous cyberattacks — on the very day of its announcement, with a private group reportedly accessing and using it regularly, though not yet for malicious cybersecurity purposes.
OpenAI launches its Agents SDK, a Python framework now boasting 24,800+ GitHub stars, enabling AI to execute complex multi-step tasks across 100+ LLMs, with a new Sandbox feature allowing agents to inspect files, run commands, and manage entire workspace environments autonomously.