#Guardrails

Meta’s AI safety system defeated by the space bar

by

July 29, 2024 at 05:09PM Meta’s machine-learning model designed to detect prompt injection attacks, known as Prompt-Guard-86M, has ironically been found vulnerable to such attacks. This model, introduced by Meta in conjunction with its Llama 3.1 generative model, aims to catch problematic inputs for AI models. However, a recent discovery by bug hunter Aman Priyanshu … Read more

How’s Uncle Sam getting on with Biden’s AI exec order? Pretty good, we’re told

by

May 27, 2024 at 12:59PM President Biden’s October executive order on AI focused on guiding federal agencies in the safe use of machine learning technologies. The order’s requirements have been well-implemented, with agencies appointing chief data officers and devising data plans. The executive order serves as guardrails for AI deployment, ensuring safe and reliable systems … Read more