In today’s digital age, people of all ages are using AI. Whether it concerns academic questions or health-related issues, many now turn to AI for immediate answers. As the use of AI grows, so do concerns regarding the online safety of children. In response, OpenAI—the company behind AI technologies—has introduced new security features for ChatGPT. These features will facilitate the identification of explicit imagery and objectionable content involving children. To this end, the company has developed a new AI-based image classifier capable of detecting such content, including images generated by AI itself.
According to OpenAI, ChatGPT will now display alerts when prompts are used in an attempt to generate content that depicts children in an inappropriate or harmful manner. Additionally, new tools and guidelines have been released for developers and law enforcement agencies.
How the new system works
OpenAI’s new image classifier scans images to identify explicit content involving children, including material generated by AI. Each image is assigned a score, which helps safety teams prioritize which images require further investigation.
This tool was developed in collaboration with the National Center for Missing & Exploited Children (NCMEC). Initial testing showed that the system reduced the number of images flagged for review by 48% compared to previous methods, while maintaining its effectiveness in identifying suspicious content.
Currently, the tool is available to select organizations via an API—including groups like the Tech Coalition and ROOST—at no additional cost.
Alerts for inappropriate prompts
OpenAI is also integrating alert screens into ChatGPT. If a user attempts to generate explicit content involving children, an alert will appear on their screen. Additionally, users will be advised to refrain from such activities and provided with information on how to seek necessary assistance. In such instances, the user's account will also be suspended for 24 hours. These alert messages were developed with the support of 'Protect Children,' the 'Lucy Faithfull Foundation,' and Joel Scanlan, a researcher from the University of Tasmania.
New tools for developers and law enforcement
OpenAI has also issued new guidelines for developers, enabling them to identify and report such content on their platforms. Meanwhile, law enforcement agencies will be provided with AI-based tools, plugins to assist in investigations, and data analysis capabilities. The company will also offer training and usage guides to investigative officials.
It is worth noting that OpenAI has been under significant pressure regarding child safety for some time. Several attorneys general from various U.S. states have even filed lawsuits against ChatGPT, alleging that it has a detrimental impact on young people. Consequently, the company is now taking a series of increasingly strict measures to prioritize safety.
Disclaimer: This content has been sourced and edited from Money Control. While we have made modifications for clarity and presentation, the original content belongs to its respective authors and website. We do not claim ownership of the content.
Contact to : xlf550402@gmail.com
Copyright © boyuanhulian 2020 - 2023. All Right Reserved.