AI News: May 31, 2026 - OpenAI Launches Biodefense AI, Codex Pro Pricing Changes Today, Build Starts Monday

1. OpenAI Launches Rosalind Biodefense - AI for Pandemic Preparedness

OpenAI announced Rosalind Biodefense on May 30, 2026 - a new initiative giving vetted developers and government partners expanded access to GPT-Rosalind, OpenAI's biology-specialized frontier model, for building biodefense and pandemic preparedness applications. The initiative is named after Rosalind Franklin, the crystallographer whose X-ray diffraction work was foundational to the discovery of DNA's double helix structure.

Access is restricted by design. OpenAI worked with external experts and public-sector partners to define the trusted access model, including the US Center for AI Standards and Innovation (CAISI), the UK AI Security Institute (UK AISI), Los Alamos National Laboratory, and the Frontier Model Forum. The rationale is explicit: biology-specialized AI capable of accelerating drug discovery and pathogen analysis carries dual-use risk - the same capabilities that help defenders detect and respond to biological threats could theoretically assist actors trying to create them. OpenAI's framing is "defensive acceleration" - ensuring frontier AI advantages defenders more than it advantages adversaries.

Approved use cases include developing new countermeasures against emerging biological threats, building early warning and detection tools for pandemic preparedness, and supporting public health organizations with disease surveillance and response. Government partners with explicit public health and biodefense missions can apply for trusted access separately from the developer program. OpenAI has not disclosed the application process publicly - organizations interested in Rosalind Biodefense access are directed to contact OpenAI directly through official government and research partnership channels.

The announcement is notable in context alongside Anthropic's Project Glasswing (Claude Mythos restricted access for cybersecurity) and Google DeepMind's AlphaGenome (genome analysis published in Nature, January 2026). All three frontier AI labs are now running restricted access programs for the highest-risk application domains - cybersecurity and biology - where the dual-use risk of unrestricted access outweighs the benefit of open availability.

2. Codex Pro 2x Promo Expires Today - What Changes from June 1

The introductory 2x capacity multiplier on the Codex Pro plan ($100/month) expires on May 31, 2026. From June 1, Codex Pro subscribers will see their effective monthly Codex capacity halve unless OpenAI announces a formal extension. This is the third introductory promo expiry in the past 10 days: Cursor's Composer 2.5 first-week 2x rate ended May 25, and Opus 4.7's GitHub Copilot promo already ended April 30. GitHub Copilot itself transitions to usage-based billing on June 1, with no published per-credit dollar price at the time of this writing.

Introductory pricing has become the industry default for major AI tool launches in 2026 - every major release since Opus 4.7 in April has launched with a time-limited multiplier. The pattern is worth documenting: teams that build workflows during the promo window at subsidized capacity and then face 2-3x rate increases within 90 days of launch are being caught off guard. If you are on Codex Pro and have not modelled your usage at standard capacity rates, June 1 is the practical deadline to do so.

For context on which plan suits your workload: ChatGPT Plus ($20/month) gives standard Codex access with the lower usage quota. The Codex add-on gives Plus users additional Codex capacity without upgrading to Pro. Codex Pro ($100/month) was designed for heavy daily Codex users - developers running multiple parallel agents, large codebase refactors, and long Goal mode sessions. If you are on Pro primarily for the 2x promo capacity, the post-June 1 standard rate is the figure to model against your actual daily usage before deciding whether to stay on Pro or step down.

3. Microsoft Build June 2-3 - Windows Becomes the AI Agent Platform

Microsoft Build 2026 runs June 2-3 at Fort Mason Center in San Francisco with a simultaneous global online stream. Satya Nadella delivers the opening keynote. Pre-event communications and confirmed session catalog details have already revealed the core announcements coming Monday, making this the most pre-announced Build in recent memory.

The strategic framing is consistent across all pre-Build communications: Microsoft is positioning Windows as the premier platform for AI agents in the same way it positioned Windows as the premier platform for enterprise software in the 1990s. The Windows Agent Runtime preview - limited to text-based agents initially - will be available to Windows Insiders in June 2026. The Azure Agent Mesh and Windows Agent Store are both targeting Q4 2026 GA.

WSL 3 is the announcement most likely to have immediate practical impact for developers. By giving Linux workloads direct access to the Windows GPU and NPU at near-native speed, it effectively closes the gap between running AI development environments on Linux and on Windows - a gap that has driven many AI developers to Mac or Linux as their primary development OS. If WSL 3 delivers on the benchmark previews (Snapdragon X Elite 2 and AMD Ryzen AI both handling multi-agent simulations 40% faster on Arm than x86 equivalents), it changes the hardware calculus for AI developers shopping for a new machine in H2 2026.

AIToolsRecap will cover the Build keynote announcements in full on June 2. Watch the May 2026 hub for the daily digest update.