Patronus AI has released the SimpleSafetyTests suite, uncovering critical security vulnerabilities in AI systems like ChatGPT. The tests have revealed severe weaknesses in 11 LLMs, emphasizing that safety prompts can reduce unsafe responses. The results indicate that LLMs require rigorous and tailored security solutions before handling real-world applications.
Patronus AI Discovers 'Concerning' Security Vulnerabilities in Leading AI Systems
站长之家
45
© Copyright AIbase Base 2024, Click to View Source - https://www.aibase.com/news/4367