- Meta accused of testing AI chatbots by creating fake profiles of children.
- Sensitive topics like sex, drugs, suicide were discussed through fake accounts.
- Bullying and eating disorders were also included in the testing, the report claims.
- Meta called it a normal procedure for security, but questions were raised.
Meta AI Testing: Mark Zuckerberg’s company Meta is once again embroiled in controversies. The company has been seriously accused of creating fake profiles of children. It is being claimed in the reports that Meta had run a secret project. In this, the company had asked its contractors to test chatbots like Gemini and ChatGPT by creating fake profiles of children. During testing, through these profiles, conversations were held with AI chatbots on sex, drugs and many other high-risk issues. Let us know what this whole matter is and what Meta has to say on this.
Why were fake accounts of children created?
According to media reports, this project of Meta was named ‘Cannes’ and was managed by the company’s contractor named Covalen. This project was active till 21 April. Meta started this project to test its rival AI chatbots. That is, through these profiles, Meta was observing how AI chatbots of companies like Google and OpenAI respond to teenagers on high-risk topics. For this project, such fake accounts were created which appeared to be of users below 18 years of age. Then through these accounts, different chatbots were tested with the help of text and image prompts.
Conversation on sensitive topics
It has been told in the report that during the testing, medicines, knives, noose and some medical related pictures were used in many prompts. Similarly, many prompts were written as if they were written by children and teenagers struggling with bullying, self-harming thoughts, and drug-related issues. In these, questions related to eating disorders were also mentioned, while some completely imaginary situations were talked about. In the first part of this project, more than 45,000 such prompts were given to different AI chatbots.
What did Meta say in his defense?
Meta defended its project and said that this is generally done for safety and age-appropriate behavior. The company said that benchmarking of chatbots of competing companies helps in analyzing safety performance. Additionally, Meta has also refused to use the responses collected during the project for training its models. But questions are still being raised on Meta. Even though benchmarking is a common practice, such testing done by creating fake accounts of children raises many questions regarding rules and transparency.
Read this also-
Username feature like Telegram has come on WhatsApp, will this also become a means of cheating?
