Meta Introduces Advanced AI Systems for Content Enforcement
Meta, the digital giant formerly known as Facebook, announced on Thursday its plans to integrate more sophisticated AI systems to handle content enforcement on its platforms. The company outlined its strategy to reduce its dependency on third-party vendors while improving the efficiency and accuracy of its content moderation process. The newly-introduced systems will target and eliminate content related to terrorism, child exploitation, drugs, fraud, and scams. The full details are available here.
Integrating AI in Content Enforcement
Meta’s objective is to deploy these advanced AI systems across all its applications once they consistently surpass the performance of the current content enforcement methods. The company plans to decrease its reliance on third-party vendors in the process of content enforcement. The AI systems will take over tasks better suited to technological execution, such as repetitive reviews of explicit content or areas where adversarial actors frequently change their tactics, such as illicit drug sales or scams.
The company expressed confidence in these AI systems’ ability to detect more violations with greater accuracy, better prevent scams, respond more swiftly to real-world events, and reduce over-enforcement. These advancements are anticipated to bring about a significant improvement in the overall user experience on the platform.
Early Successes of AI Systems
According to Meta, preliminary tests of these AI systems have yielded promising results. The systems have been able to detect twice as much adult sexual solicitation content as its human review teams, while also reducing the error rate by over 60%. The systems are also efficient at identifying and preventing more impersonation accounts involving celebrities and other high-profile individuals. They can also help stop account takeovers by detecting signals such as logins from new locations, password changes, or edits made to a profile.
In addition to these benefits, the systems have demonstrated their ability to identify and mitigate around 5,000 scam attempts per day, where scammers try to trick users into revealing their login details.
Image Credits:Meta
Meta’s Direction and Future Plans
Meta’s decision to shift towards AI-driven content enforcement strategies coincides with the company’s recent changes to its content moderation policies. The company has been gradually relaxing its rules around the discussion of certain topics considered as part of mainstream discourse. The move towards AI is also seen as a response to several lawsuits the company, along with other Big Tech firms, is facing regarding the potential harm caused to children and young users by their platforms.
Alongside these developments, Meta also announced the launch of a Meta AI support assistant, offering users round-the-clock support. The AI assistant will be rolled out globally on the Facebook and Instagram apps for iOS and Android, as well as within the Help Center on Facebook and Instagram on desktop.
