Today we’re covering the biggest AI stories of April 17th, 2026. OpenAI dropped a major Codex update that adds background computer use, parallel agents, an in-app browser, memory, image generation, and automations — with Codex’s head openly calling it “building the superapp out in the open” as the platform evolves from a coding agent into a unified AI command center. Anthropic released Claude Opus 4.7, its new top publicly available model that jumps ahead of GPT-5.4 and Gemini 3.1 Pro on agentic coding benchmarks — while still trailing the company’s own locked-away Mythos Preview, marking the first time the public frontier feels deliberately behind what Anthropic is keeping for itself. Plus, OpenAI launched GPT-Rosalind, its first life sciences domain model built for drug discovery and biological research, scoring better than 95% of human scientists on a blind RNA prediction test — and a community workflow from a 73-year-old author who uses Claude to design book covers, format interiors, market to book clubs, and run his entire social media presence.