According to the official announcement from OpenAI on May 15, to improve transparency in model security, OpenAI has launched the “Safety Evaluation Center”. This center continuously publishes the safety performance results of models regarding harmful content, jailbreak attacks, hallucination generation, and prioritizes guidance. Unlike system cards that only disclose data once upon model release, the center will be updated periodically alongside model updates, supporting horizontal comparisons between different models. The goal is to enhance the community's understanding of AI safety and transparency in governance. Currently, GPT-4.5 and GPT-4o excel in their resistance to jailbreak attacks and real-world accuracy.
This content is for informational purposes only and does not constitute investment advice.
Source: https://tintucbitcoin.com/openai-trinh-lang-trung-tam-danh-gia-an-toan-moi/
Thank you for reading this article!
Please Like, Comment, and Follow TinTucBitcoin to always stay updated with the latest news about the cryptocurrency market and not miss any important information!