// Today in AI

Today in AI: Policy Shifts and Product Launches

A major regulatory hurdle for Anthropic was cleared, allowing its most advanced models to return globally. Meanwhile, the industry focused on more practica

LDLatentDaily Desk Jul 1, 2026 2 min read

Today in AI — Photo by Kindel Media on Pexels

July 01, 2026

A major regulatory hurdle for Anthropic was cleared, allowing its most advanced models to return globally. Meanwhile, the industry focused on more practical, cost-effective offerings, from cheaper image models to new agent benchmarks.

⚖️ US Lifts Export Controls on Anthropic's Top Models

The Department of Commerce removed restrictions on Claude Fable 5 and Mythos 5, allowing Anthropic to restore global access. This resolves a major, disruptive policy intervention that had sidelined two of the world's most capable models for weeks, providing much-needed clarity to the industry. (@AnthropicAI)

🚀 Anthropic Launches Claude Science for Research

Anthropic announced Claude Science, a new flagship product designed to autonomously conduct and support scientific research. This follows the playbook of Claude Code, aiming to create a specialized, high-value assistant for a critical professional domain. (MIT Tech Review)

🧠 Google Releases Faster, Cheaper Image Model

Google DeepMind launched Nano Banana 2 Lite, billing it as its fastest and cheapest Gemini image model. This is a clear move to compete on price and speed in the increasingly commoditized text-to-image space, prioritizing utility over maximum fidelity. (@GoogleDeepMind)

🔬 OpenAI Introduces GeneBench-Pro for Bio Agents

OpenAI released GeneBench-Pro, a research-level benchmark designed to test AI agents on messy, judgment-heavy biological data analysis. This signals a push beyond simple task completion towards evaluating AI's ability to navigate complex, real-world research workflows. (@OpenAI)

💰 Wayve's $8.5B Valuation in Employee Tender Offer

The AV startup Wayve launched an $85M employee tender offer at an $8.5B valuation. This highlights how leading AI companies are using liquidity events as a strategic tool to attract and retain top talent in a fiercely competitive market. (TechCrunch)

🔬 New OS and Enterprise Agent Benchmarks Emerge

Two new benchmarks were released: OSWorld2.0 for long-horizon computer use agents and ScarfBench for enterprise Java framework migration. The focus is squarely on creating rigorous tests for practical, autonomous agent performance in real-world environments. (@_akhaliq)

The takeaway: **The sudden lifting of export controls on Anthropic's models removes a major source of uncertainty, allowing the company to fully re-engage in the global model race.**