Diving Deeper into AI Package Hallucinations
www.lasso.security
Lass Security's recent research on AI Package Hallucinations extends the attack technique to GPT-3.5-Turbo, GPT-4, Gemini Pro (Bard), and Coral (Cohere).

LLM responds with fake name of packages. Real package made using the fake name and gets installed 30,000 times in 3 months. Malicious actors could use this behavior.