In their own words.
Our plain version above is ours. This is the official or expert wording we based it on.
“a structured testing effort to find flaws and vulnerabilities in an AI system,”
Old (2023) definition; still the common meaning.
Related words.
Benchmark
A fixed test, such as a set of exam questions or coding tasks, used to score and compare AI models. Scores can mislead if models were trained on the answers or the test is too easy.
Deepfake
Fake but realistic video, audio or images of real people, places or events, made or altered with AI.
Alignment
Work to make AI systems pursue the goals and values their makers and users intend, and not harmful or unintended ones.
Jailbreak
A prompt crafted to trick a model into ignoring its safety rules, for example to get instructions it is meant to refuse.
Prompt injection
An attack in which hidden or malicious instructions, typed in or planted in a web page, email or file the AI reads, make it do something its user did not intend.
System card (model card)
A report a lab publishes with a model, describing what it can do, how it was tested, safety results and known limits. The idea comes from 'model cards'.