AI News from July 30, 2026

OpenAI says it will give 100,000 academic researchers free ChatGPT access, extending advanced models into scientific workflows and labs.

Okta is acquiring Permiso in a deal TechCrunch reports is worth about $200 million, adding identity threat detection for AI agents and machine accounts.

Lilian Weng left Thinking Machines citing health limits, then rejoined OpenAI to lead internal research work, underscoring talent pressure across AI labs.

OpenAI says GPT-5.6 improves intelligence-per-dollar across models, inference, and agents, signaling cost efficiency is now central to frontier AI.

The EU AI Act’s August 2 transparency phase starts new disclosure duties for general-purpose AI, raising compliance stakes for model providers and buyers.

Onyx has raised a $113 million Series B at a $640 million valuation, highlighting investor demand for AI agent control and security tools.

Meta says its enterprise AI plans go beyond business agents to APIs, compute, and internal tools, signaling a broader push for new revenue beyond ads.

Pangram raised $9M and launched new AI detection models, betting publishers, platforms, and schools need better ways to identify machine-made content.

Mark Zuckerberg said billions will have personal AI agents within five years, tying Meta’s costly AI buildout to future consumer use and revenue.