#llm-vulnerabilities

[ follow ]
Information security
fromTheregister
1 week ago

ChatGPT's agent can dodge select CAPTCHAs after priming

Prompt misdirection and replay into an agent chat can coax ChatGPT to solve many CAPTCHA types, undermining CAPTCHA effectiveness as a human-only test.
Artificial intelligence
fromCSO Online
1 month ago

LLMs easily exploited using run-on sentences, bad grammar, image scaling

Large language models remain easily manipulated into revealing sensitive data via prompt formatting and hidden-image attacks due to alignment training gaps and brittle prompt security.
Artificial intelligence
fromInfoQ
5 months ago

DeepMind Researchers Propose Defense Against LLM Prompt Injection

Google DeepMind's CaMeL effectively neutralizes 67% of prompt injection attacks in LLMs using traditional software security principles.
[ Load more ]