A term that we should all be learning is Prompt Injection. Research papers from 14 academic institutions in 8 countries contained hidden directives to artificial intelligence to only provide positive information. These were then hidden by white text or small font sizes, but then passed along to AI which interprets all text, regardless of visibility to us. This trick isn’t old, and has been used by spammers, people on job websites, and more. The name itself, is newer, and refers to AI prompts or requests. Clearly, this will be increasingly used more as those with loose morals or scruples find out about it.
TOKYO — Research papers from 14 academic institutions in eight countries — including Japan, South Korea and China — contained hidden prompts direct
Found at asia.nikkei.com
