Meta has acknowledged a technical glitch that led to the erroneous suspension of several Facebook Groups, but insists there is no widespread problem affecting its platforms.
Group administrators have reported receiving automated messages falsely claiming their groups violated Meta’s policies, resulting in deletions. A notable example includes a 680,000-member group focused on bug memes, which was removed after being flagged for breaching rules on "dangerous organisations or individuals." The group has since been reinstated. Another administrator, managing a 3.5 million-member AI-focused group, reported a temporary suspension, with Meta later admitting its technology "made a mistake."
The issue has sparked broader concerns, with some Instagram users reporting similar account suspensions. Many users attribute these errors to Meta’s AI-driven moderation tools. A petition on Change.org, titled “Meta wrongfully disabling accounts with no human customer support,” has amassed nearly 22,000 signatures, highlighting frustrations over inaccessible human support. Reddit threads also feature accounts of users losing access to sentimental or business-related pages, with some claiming bans for alleged breaches of policies on child sexual exploitation.
Meta, however, denies a surge in erroneous suspensions. A spokesperson told the BBC, “We take action on accounts that violate our policies, and people can appeal if they think we’ve made a mistake.” The company employs a mix of AI and human reviewers to enforce its Community Standards, with AI often flagging content before human review in certain cases. Meta’s latest Community Standards Enforcement Report (January–March 2025) recorded 4.6 million actions on child sexual exploitation content, the lowest since early 2021.
The company also highlighted its use of technology to detect suspicious behaviour, such as adults repeatedly searching harmful terms or being reported by teen accounts, which could result in restrictions or account removals. Meta’s next transparency report is expected in the coming months.
The issue has fuelled ongoing debates about the reliability of AI moderation and the need for accessible human support for affected users.
