The Most Important Biology Release of the Year, OpenAI Funds Its Own Fallout, AI Gets Self-Improving

Today we're covering the biggest AI stories of May 28th, 2026. Mark Zuckerberg and Priscilla Chan's Biohub released Evolutionary Scale Models — an open-source protein world model trained on 2.8 billion sequences that predicts structure, designs novel proteins, and already hit 36 to 88% success rates against five cancer and immune disease targets, outperforming AlphaFold and putting drug discovery infrastructure directly in the hands of researchers everywhere. OpenAI Foundation committed $250 million to help workers, communities, and economies navigate AI-driven disruption — funding efforts to track how AI value actually flows, support job transitions, and explore tax shifts from labor to capital, even as layoffs tied to AI efficiency continue spreading across industries. Plus, a new startup called Trajectory launched with $15 million to build AI that keeps learning from real-world corrections and user edits after deployment — post-training models every week and targeting hourly updates — and today's community workflow comes from Michel in Solana Beach, who used Claude, Cursor, and Claude Code to build a one-click inventory verification app for his wine distribution company that instantly tells salespeople what's in stock, ending the interruptions and speeding up sales.