Meta Deploys Advanced AI Systems for Content Enforcement, Cutting Third-Party Vendor Reliance
Meta says its new AI enforcement systems detect twice the adult sexual solicitation content and reduce error rates by more than 60 percent compared to human review teams, while flagging around 5,000 scam attempts daily.
Meta announced Thursday that it is beginning to roll out more advanced AI systems to manage content enforcement across its platforms, reducing its reliance on third-party vendors in the process. The company said the new AI systems are designed to handle the enforcement of policies around terrorism, child exploitation, drugs, fraud, and scams — categories that previously required extensive human review and contractor labor.
The systems will be deployed across Facebook, Instagram, and other Meta properties once they consistently outperform existing enforcement methods in internal testing.
The performance figures Meta shared are striking. In early tests, the new AI systems identified twice as much adult sexual solicitation content as human review teams while reducing the error rate by more than 60 percent.
The systems can also detect and prevent more impersonation accounts targeting celebrities and high-profile individuals, identify account takeover attempts by flagging anomalous login patterns or profile edits, and mitigate approximately 5,000 scam attempts per day in which bad actors try to steal user credentials.
Meta said human experts will continue to design, train, and oversee the AI systems and will remain responsible for the most complex or high-impact decisions — including appeals of account disablement and responses to law enforcement requests. The company believes AI is better suited than human reviewers for repetitive tasks involving graphic content or rapidly evolving adversarial tactics like drug sale promotions and financial fraud schemes, where consistency and speed are more important than case-by-case judgment.
Alongside the content enforcement announcement, Meta launched a Meta AI support assistant providing 24-hour access to automated help through the Facebook and Instagram apps on iOS and Android, as well as through the Help Center on desktop. The dual announcements reflect Meta's broader strategic push to replace human-intensive operations with AI systems as the company continues to invest heavily in AI infrastructure and reduce operational costs across its platforms.
Read the original reporting at TechCrunch.
