Researchers get AI ‘drunk’ to expose new cyber security risks in chatbots
By Kyle Mackey-Laws Fine-tuning large language models using drunken text makes them more likely to leak secrets and answer harmful questions, new UNSW-led research shows. UNSW researchers have shown that large language models (LLMs) can be pushed to imitate drunken behaviour – and that once they do…
Read the full story at AIhub ↗
Timeline · 1 report
- 2026-10-27 12:14 · AIhub
Researchers get AI ‘drunk’ to expose new cyber security risks in chatbots