AI biology is still early enough for its safeguards to be built into the stack. @RANDCorporation maps nine interventions across access to models and data, material screening, detection, attribution and public-health response. No single guardrail can cover a chain running from information to execution. BIOS ( @bioaidevs ) makes the timing concrete. Several stages of computational drug design can already run inside one campaign. Wet-lab proof is the next step. Build the safeguards before the loop closes.
