From anonymization algorithms to zero-knowledge proofs, our no-nonsense glossary unpacks the encryption keys, clean rooms and ...

Lenovo's AI chatbot, Lena, has critical security flaws allowing attackers to run malicious code, steal session cookies, and potentially access corporate systems.

随着大语言模型在各行各业的广泛落地，幻觉问题已成为影响其可靠性和实际价值的核心挑战。本文系统分析了LLM幻觉产生的原因，评估了其优缺点，并从Prompt设计、RAG检索增强、采样参数、SFT监督微调、强化学习、思维链推理、Agent ...