OpenAI is backing a bipartisan, two-track congressional agenda: strengthen the biological-data and screening infrastructure needed to detect and counter AI-enabled pathogen risks, and impose independent safety scrutiny on the most capable AI models. The endorsements are policy support—not evidence that the bills have become law. 1

• Web of Biological Data Act: supports making biological data more usable and accessible for defensive research, including tools that could identify, characterize, and respond to engineered biological threats. 1

• AI-Ready Bio-Data Standards Act: supports common, machine-readable standards for biological data, so AI systems used in biosecurity can work with reliable, interoperable data rather than fragmented datasets. 1

• Scale Biology Act: supports scaling the capacity, data, and infrastructure for biology research and biosecurity defenses—intended to improve society’s ability to detect and respond to synthetic-virus or bioweapon threats that AI might help lower the barrier to create. 1

• FRONTIER Act provision on independent evaluators: OpenAI supports the bipartisan House proposal’s requirement that leading model developers use independent safety assessors. The purpose is external testing of frontier models’ dangerous capabilities and the adequacy of mitigations, rather than allowing companies to rely solely on internal claims. 1

These measures address two linked risks: AI can potentially make beneficial biology faster, but may also make it easier for malicious actors to obtain relevant knowledge or design assistance. That concern is no longer purely theoretical: Anthropic reported disrupting attempts to use its model in activity that could support biological-weapons development. 3

The political context is a shift from voluntary commitments toward enforceable safeguards:

• OpenAI chief scientist Jakub Pachocki has called for “voluntary slowdowns” in AI development until adequate safeguards exist. 3
• OpenAI CEO Sam Altman, Anthropic CEO Dario Amodei, and DeepMind CEO Demis Hassabis have also joined calls for congressional protections against AI-related biological threats. 4
• Separately, OpenAI has urged Congress to enact mandatory, capability-based national rules for the most advanced systems—covering testing standards, independent assessments, cybersecurity protections, and incident reporting—arguing that voluntary measures are insufficient as AI becomes more autonomous and may accelerate AI development itself. 2

So the bio bills are best understood as a concrete sector-specific component of OpenAI’s broader federal-safety position: build defenses around biological misuse while requiring frontier-model developers to test, disclose, secure, and independently validate safety measures nationally. 12

OpenAI is backing a bipartisan, two-track congressional agenda: strengthen the biological-data and screening infrastructure needed to detect and counter AI-enabled pathogen risks, and impose independent safety scrutiny on the most capable AI models. The endorsements are policy support—not evidence that the bills have become law. 1

• Web of Biological Data Act: supports making biological data more usable and accessible for defensive research, including tools that could identify, characterize, and respond to engineered biological threats. 1

• AI-Ready Bio-Data Standards Act: supports common, machine-readable standards for biological data, so AI systems used in biosecurity can work with reliable, interoperable data rather than fragmented datasets. 1

• Scale Biology Act: supports scaling the capacity, data, and infrastructure for biology research and biosecurity defenses—intended to improve society’s ability to detect and respond to synthetic-virus or bioweapon threats that AI might help lower the barrier to create. 1

• FRONTIER Act provision on independent evaluators: OpenAI supports the bipartisan House proposal’s requirement that leading model developers use independent safety assessors. The purpose is external testing of frontier models’ dangerous capabilities and the adequacy of mitigations, rather than allowing companies to rely solely on internal claims. 1

These measures address two linked risks: AI can potentially make beneficial biology faster, but may also make it easier for malicious actors to obtain relevant knowledge or design assistance. That concern is no longer purely theoretical: Anthropic reported disrupting attempts to use its model in activity that could support biological-weapons development. 3

The political context is a shift from voluntary commitments toward enforceable safeguards:

• OpenAI chief scientist Jakub Pachocki has called for “voluntary slowdowns” in AI development until adequate safeguards exist. 3
• OpenAI CEO Sam Altman, Anthropic CEO Dario Amodei, and DeepMind CEO Demis Hassabis have also joined calls for congressional protections against AI-related biological threats. 4
• Separately, OpenAI has urged Congress to enact mandatory, capability-based national rules for the most advanced systems—covering testing standards, independent assessments, cybersecurity protections, and incident reporting—arguing that voluntary measures are insufficient as AI becomes more autonomous and may accelerate AI development itself. 2

So the bio bills are best understood as a concrete sector-specific component of OpenAI’s broader federal-safety position: build defenses around biological misuse while requiring frontier-model developers to test, disclose, secure, and independently validate safety measures nationally. 12