OpenAI has flagged its ChatGPT Agent as a high-risk AI tool, capable of creating dangerous biological or chemical substances. This advanced system can perform various tasks like scheduling meetings, browsing websites, analyzing data, and automating complex workflows. It merges features from previous OpenAI tools, offering users seamless task execution while maintaining user control.
The agent utilizes powerful tools, such as visual and text-based browsers, to execute commands and connect with apps like Gmail. In tests, ChatGPT Agent has outperformed previous models on benchmarks for reasoning and data analysis. However, it raises significant privacy and security concerns due to the extensive permissions required.
Industry experts warn that the rapid development of AI agents outpaces public understanding, increasing risks of data leaks and misuse. To mitigate these dangers, it’s recommended that companies adopt a “human-in-the-loop” approach, emphasizing supervision and informed use. This ensures safety while leveraging AI advancements effectively.
Source link
