Sign in Subscribe

model-vulnerabilities

Aug
04
Privacy Backdoors: Stealing Data with Corrupted Pretrained Models (Paper Explained)

Privacy Backdoors: Stealing Data with Corrupted Pretrained Models (Paper Explained)

🆕 from Yannic Kilcher! Learn how attackers can steal data from AI models by manipulating weights, posing serious privacy risks. #AI
4 min read
Apr
07
"Many Shot" Jailbreak - The Bigger the Model, The Harder it Falls

"Many Shot" Jailbreak - The Bigger the Model, The Harder it Falls

🆕 from Matthew Berman! Discover the risks of Many Shot Jailbreaking in AI models - exploiting large context windows for harmful
2 min read
Mar
09
NEW AI Jailbreak Method SHATTERS GPT4, Claude, Gemini, LLaMA

NEW AI Jailbreak Method SHATTERS GPT4, Claude, Gemini, LLaMA

🆕 from Matthew Berman! Discover how ASKII art-based jailbreaks challenge top language models. Uncover vulnerabilities in GPT-4, Claude, Gemini, and LLaMA.
3 min read