AI Can Models Creata Backdoors, Research Says

Bad samples can poison any AI model, study finds

A new study by Anthropic finds that just 250 harmful files can poison even the largest AI models, showing how easily training data can be corrupted. The post Bad samples can poison any AI model, study finds appeared first on CoinGeek.

AI Can Models Creata Backdoors, Research Says

Scraping the internet for AI training data has limitations. Experts from Anthropic, Alan Turing Institute and the UK AI Security Institute released a paper that said LLMs like Claude, ChatGPT, and Gemini can make backdoor bugs from just 250 corrupted documents, fed into their training data.  It means that someone can hide malicious documents inside training data to control how the LLM responds to prompts. About the research  It trained AI LLMs r…