Meta admitted that it used public posts on its Facebook and Instagram platforms to train a new artificial intelligence virtual assistant.
According to a Reuters report, Nick Clegg, Meta’s president of global affairs, said the company only uses public posts, including private posts to share with friends and family and non-public posts to train its AI bots. He said he had avoided both public messages.
“We have sought to exclude data sets that are predominantly personal information,” Clegg said at the company’s annual Connect conference, adding that “the vast majority” of the data used was already publicly available.
NSA announces new artificial intelligence security center: ‘Sorely needed’
Tech companies have come under fire in recent months over reports that they use information from the internet with permission to train AI models that can sort through large amounts of data. According to Reuters, data wipes can lead to lawsuits in some cases, especially if the AI is accused of copying copyrighted material.
“AI requires vast amounts of training data, so user submissions are an ideal way to ‘feed the beast,'” said Christopher Alexander, chief analytics officer at Pioneer Development Group on FOX. He told News Digital. “The concern is how these better-trained AI personas will be leveraged. There is some real concern about how human-like AI will appear and that needs to be taken into account. ”
What is artificial intelligence (AI)?
John Schweppe, policy director for the American Principles Project, questioned how many “safeguards” are in place to protect personal information.
“Meta appears to be building its AI based on user posts, photos, and personal data. Did consumers really sign up for that?” Schweppp told Fox News Digital. . “America is at a flashpoint, similar to the situation it was in nearly 20 years ago when social media emerged and launched the greatest social experiment in human history. Will Congress act now and give Americans oversight over AI? , or we will deal with it.” We will once again be at the mercy of our technological masters. ”
Warren blows up Senate AI closed session, calls for swift regulation
According to Reuters, Meta CEO Mark Zuckerberg publicly announced the company’s new AI tool at a product conference last week, which is similar to Llama 2’s large-scale language model. It is said to have been made using a custom model. The product can generate text, audio, and images, while integrating with Bing search to access real-time information, the report said.
Public posts on Facebook and Instagram will be used to train Meta’s AI tools for both image generation and chat responses, and user interactions with the bot will help improve future features, Meta told Reuters.
Given Heavens, director of policy at the Bull Moose Project, told Fox News Digital that it “doesn’t come as a surprise to users” that their posts will be used to train Meta’s AI tools, but that users ” You should be concerned about whether your post is ‘your post’ Data is used in a responsible and safe manner. ”
Click here for more US news
“Without real action from Congress, Americans will be forced to assume that these AI companies are responsible for their data, but given the past decade, many Americans are unlikely to believe that.” It’s going to be tough,” Havens said. “If Congress does not act, data privacy concerns will only increase.”
Phil Siegel, founder of the Center for Advanced Preparedness and Threat Response Simulation, told Fox News Digital that it’s “not surprising” that Meta is using posts to train its AI, adding that it is “not surprising” that Meta is using posts to train its AI. The goal is to train an LLM (Large-Scale Language Model) to behave as if it were a social media user, allowing access to unique data and allowing it to access facts from the Internet world. This is distinct from an LLM that only collects information. ”
But Siegel noted that there could be concerns about the prevalence of artificial personalities, especially given the mental health effects social media already has on users.
CLICK HERE TO GET THE FOX NEWS APP
“Models spreading bad or unpleasant information, excessive hallucinations (think of combining human emotional responses with LLMs that are already hallucinating), and teenagers “I’m worried that we might create distorted personalities to interact with. This might actually be the way the metaverse works,” he said. He said. “To be honest, social media companies have harmed the mental health of many people, young and old, without the involvement of AI, and we need to ensure that this AI does not exacerbate the problem.”
Mehta did not immediately respond to Fox News Digital’s request for comment.