A New Attack on ChatGPT—and No One Knows How to Stop It

Adversarial attacks exploit the way that machine learning picks up on patterns in data to produce aberrant behaviors.

These language models are also prone to fabricating information, repeating social biases, and producing strange responses as answers prove more difficult to predict.

Armando Solar-Lezama, a professor in the MIT’s college of computing, says it makes sense that adversarial attacks exist in language models, given that they affect many other machine learning models.

But these language models do not have the same kinds of problems because they do not rely on similar corpora of text data.

Each company introduced blocks to prevent the exploits described in the research paper, but they have not figured out how to block adversarial actions more generallyMore generally, solar-lezama suggests, one reason for this may be that all large language model are trained on similar datasets, much of it downloaded from the same websites

#shorts #techshorts #technews #tech #technology #models #adversarial attacks #research

👋 Feeling the vibes?

Keep the good energy going by checking out my Amazon affiliate link for some cool finds! 🛍️

If not, consider contributing to my caffeine supply at Buy Me a Coffee ☕️.

Your clicks = cosmic support for more awesome content! 🚀🌈