OpenAI launches Rosalind biodefense and governance framework as Trump kills AI safety order

On May 29, OpenAI announced the Rosalind Biodefense initiative, extending access to its top life-sciences model GPT-Rosalind to vetted developers and U.S. government and allied partners for defensive applications like epidemiological modeling and vaccine development. Partners include Lawrence Livermore National Lab, Johns Hopkins APL, and CEPI. The trusted-access model restricts distribution to prevent misuse, signaling a shift from fully public to conditional release for frontier AI models in biology. That same day, OpenAI published its Frontier Governance Framework, mapping safety practices to California's Transparency in Frontier AI Act and the EU's GPAI Code of Practice. The framework quantitatively defines systemic risk as more than 50 deaths or $1 billion in property damage from a single event, with threat domains including cyber, CBRN, manipulation, and loss of control.

Just over a week earlier, on May 21, President Trump withdrew a draft executive order that would have established a voluntary 90-day pre-release safety review for frontier AI models. According to Axios and Semafor, Elon Musk, Mark Zuckerberg, and David Sacks each called Trump directly on May 20-21 to push for the withdrawal, with Trump reportedly saying, 'We're beating China, and I didn't want to get in the way of that lead.' The contrast is stark: while government pulls back even voluntary oversight, OpenAI proactively publishes its own self-governance framework. This reinforces a growing trend where corporate self-regulation fills the void left by retreating public policy, but raises questions about enforcement and trust without independent oversight.