-Systems using artificial intelligence might be capable of performing +Systems using what is today commonly referred to as "artificial intelligence" and/or "machine learning" might be capable of performing complex tasks at a level that meets or exceeds human performance. This might include tasks such as the acquisition and use of [=verifiable credentials=]. -Therefore, using such tasks to determine a distinction between human activity and an automated "bot", -as commonly done with a CAPTCHA today, -might no longer provide adequate protections. +Using such tasks to distinguish between human and automated "bot" activity, as is +commonly done today with a CAPTCHA, +for instance, might thereby cease to provide adequate or acceptable protection.
Implementers of security architectures that use [=verifiable credentials=] -and perform validation are urged to consider artificial intelligence actors that -might legitimately hold [=verifiable credentials=] to interact with other -systems. Implementers might also consider how threat actors could couple -artificial intelligence with [=verifiable credentials=] to pose as humans when -interacting with their systems. Such systems include, but are not limited to, -global infrastructure such as social media, election, energy distribution, -supply chain, and autonomous vehicle systems. +and/or perform validation on their content are urged to consider the existence +of machine-based actors, such as those which are today commonly referred to as +"artificial intelligence", that might legitimately hold [=verifiable credentials=] +for use in interactions with other systems. Implementers might also consider how +threat actors could couple such "artificial intelligence" systems with [=verifiable +credentials=] to pose as humans when interacting with their systems. Such systems +might include, but not be limited to, global infrastructure such as social media, +election, energy distribution, supply chain, and autonomous vehicle systems.