prompt injections

Noun

prompt injection (countable and uncountable, plural prompt injections)

1. (artificial intelligence, computer security) A method of manipulating an artificial intelligence system by embedding malicious instructions (prompts) within input data in order to trick the model into overriding its intended behaviour or safeguards.
   • Yesterday, OpenAI announced an API for ChatGPT and posted an underlying format for the bot on GitHub, alluding to the issue of prompt injections.2023 March 3, Chloe Xiang, “Hackers Can Turn Bing's AI Chatbot Into a Convincing Scammer, Researchers Say”, in VICE, archived from the original on 22 Mar 2023
2. (artificial intelligence, computer security) A method of causing an artificial intelligence to ignore its initial instructions (often ethical restrictions) by directly giving it a particular prompt; a jailbreak.
   • Retomeli.io is a jobs board for remote workers, and the website runs a Twitter bot that spammed people who tweeted about remote working. The Twitter bot is explicitly labelled as being "OpenAI-driven", and within days of Goodside's proof-of-concept being published, thousands of users were throwing...2022 September 21, Alex Hern, “TechScape: AI's dark arts come into their own”, in The Guardian, London: Guardian News & Media, →ISSN, →OCLC, archived from the original on 05 Feb 2023