Final yr, we introduced our Safe AI Framework (SAIF) to assist others safely and responsibly deploy AI fashions. It not solely shares our greatest practices, however presents a framework for the business, frontline builders and safety professionals to make sure that when AI fashions are carried out, they’re safe by design. To drive the adoption of important AI safety measures, we used SAIF rules to assist kind the Coalition for Safe AI (CoSAI) with business companions. Immediately, we’re sharing a brand new device that may assist others assess their safety posture, apply these finest practices and put SAIF rules into motion.
The SAIF Threat Evaluation, out there to make use of right now on our new web site SAIF.Google, is a questionnaire-based device that can generate an on the spot and tailor-made guidelines to information practitioners to safe their AI programs. We imagine this simply accessible device fills a important hole to maneuver the AI ecosystem towards a safer future.
New SAIF Threat Evaluation
The SAIF Threat Evaluation helps flip SAIF from a conceptual framework into an actionable guidelines for practitioners liable for securing their AI programs. Practitioners can discover the device on the menu bar of the brand new SAIF.Google homepage.
