Scientists warn AI can design viruses without safety governance

Two researchers have warned that generative AI models can now compose functional viral genomes while regulatory frameworks remain insufficient to prevent misuse. The scientists stated in a new publication that the technical capability to design biological agents currently outpaces global governance mechanisms intended to ensure safe development. This disparity raises immediate concerns regarding dual-use research and biosecurity protocols within the rapidly expanding AI sector. The warning highlights specific vulnerabilities where open-weight models or accessible APIs could lower barriers to pathogen engineering for malicious actors. Industry stakeholders have yet to establish standardized screening or red-teaming requirements for biological sequence generation. The researchers argue that voluntary commitments are inadequate given the speed of model advancement. Their statement calls for binding international standards to align AI deployment with biosafety norms before capabilities become irreversible. No specific AI company was named in the alert.