Prompt injection attacks against AI - Printable Version +- Bezbedan Balkan (https://bezbedanbalkan.net) +-- Forum: Opšte teme (https://bezbedanbalkan.net/forum-1.html) +--- Forum: Vesti, zanimljivosti i razno (https://bezbedanbalkan.net/forum-26.html) +--- Thread: Prompt injection attacks against AI (/thread-152.html) |
Prompt injection attacks against AI - 1van - 10-15-2022 Pisanje eksploita prostim jezikom Quote:OpenAI’s API provides access to GPT-3, which performs a wide variety of natural language tasks, and Codex, which translates natural language to code. Problemi su opisani ovde: - https://simonwillison.net/2022/Sep/16/prompt-injection-solutions/ - https://simonwillison.net/2022/Sep/12/prompt-injection/ Quote:The more I think about these prompt injection attacks against GPT-3, the more my amusement turns to genuine concern. Quote:If I had a protection against XSS or SQL injection that worked for 99% of cases it would be only be a matter of time before someone figured out an exploit that snuck through. RE: Prompt injection attacks against AI - 1van - 02-04-2023 Zanimljiv primer: https://twitter.com/semenov_roman_/status/1621465137025613825. RE: Prompt injection attacks against AI - 1van - 02-13-2023 Još jedan zanimljivi primer: https://twitter.com/kliu128/status/1623472922374574080, i detalji: https://arstechnica.com/information-technology/2023/02/ai-powered-bing-chat-spills-its-secrets-via-prompt-injection-attack/. |