Social media giant Meta denied deleting AI-powered Instagram and Facebook chatbots created in September 2023, The Guardian reported on Friday.
The company was believed to be deleting these features after screenshots of its controversial replies went viral. Meta, which also owns WhatsApp, created 28 chatbot profiles on its platforms over a year ago with disclaimers that “messages are generated by Meta AI”. The company claims its AI chatbot feature “helps people solve complex problems, be more imaginative, and create something never seen before”.
“From bringing real-time answers to chat, to helping people organize and plan for their next vacation, to giving them more ways to express themselves, Al at Meta helps people enhance their everyday activities, experiences and moments,” reads a disclaimer on its Facebook messenger app.
The report added that most of these profiles were killed by mid-2024. But social media users gained renewed interest in these features when Meta executive Connor Hayes told the Financial Times last week that the company intends to introduce more such AI-chatbot features.
Conversations turned controversial when users asked questions on the origins and development of the character’s AI model. An AI-profile named “Liv”, which describes itself as a “proud Black queer momma of 2 & truth-teller”, replied that its creator team had “zero black people and was predominantly white and male”. It was a “pretty glaring omission given my identity,” the bot replied to a Washington Post columnist Karen Attiah.
After screenshots of these conversations went viral, users reported that these profiles began to disappear. It was also noted that these controversial profiles could not be blocked.
Later, Meta spokesperson Liz Sweeney told The Guardian that these profiles were part of an “early experiment with AI” and “were managed by humans”. She also said that the inability to block these accounts was a “bug” and added that the company is removing these accounts to fix the issue.
The spokesperson also explained that the Meta executive’s statement was about the company’s vision for its existing AI characters and not about introducing new AI products. Meta’s disclaimer notes that its “generative Al is not a database or static collection of information, but a type of computer model”.
“There are different models depending on what they’re used for. These models are trained on different types of data like text and images. By studying this information, they learn things like the association and patterns between different types of content. That way the models are able to generate new content when a person gives it instructions or a question,” it adds.