Security Level 5

Security Level 5 (SL5) is a registered 501(c)(3) nonprofit developing a security standard for frontier AI systems against nation-state-level adversaries. The organisation works with leading AI laboratories and U.S. intelligence agencies, and draws on expertise in cybersecurity, AI safety and national security. It is based in the San Francisco Bay Area. Its central output is the SL5 Standard, an integrated security control profile for frontier AI infrastructure. The profile is anchored in established frameworks such as NIST SP 800-53 and extended with supply-chain and facility requirements. Version 0.1, released in 2026, concentrates on long lead time controls - facility construction, hardware procurement and organisational capability development - that must be planned years in advance. A target SL5 security posture is set for 2028/2029. Alongside the standard, SL5 is designing architectures purpose-built for frontier AI workloads and standing up reference technology stacks to guide laboratory buildouts. The work spans infrastructure security, facility security and supply chain security, and is carried out through sustained engagement with frontier AI labs, government partners and security experts. The stated aim is to create technical and strategic optionality for frontier AI systems.