Taiwan Job Scam Warning

UPDATED MAR 30, 2026
2026-003-FB-UA
Today, October 23, 2025, the Oversight Board selected a case appealed by a Facebook user regarding a post reshared by a police department in Taiwan. The post contains an image of animated pigs and a bird in a police uniform. Overlay text and the caption, in Chinese, describe common signs of job scams and warn job seekers. The caption ends with information about an anti-scam hotline.
Upon initial review, Meta took down this content for violating our policy on Human Exploitation, as laid out in the Community Standards. However, upon additional review, we determined we removed the content in error and reinstated the post.
We will implement the Board’s decision once it has finished deliberating, and we will update this post accordingly. Please see the Board’s website for the decision when they issue it.
Read the board’s case selection summary
Case decision
We welcome the Oversight Board's decision today, January 29, 2026, on this case. The Board overturned Meta’s original decision to remove the content. Meta previously reinstated this content, and as a result, no further action will be taken on the case content.
When it is technically and operationally possible to do so, we will also take action on content that is identical and in the same context. For more information, please see our Newsroom post about how we implement the Board’s decisions.
After conducting a review of the recommendation provided by the Board, we will update this post with initial responses to that recommendation.
Recommendations
Recommendation 1 (implementing in part)
To disrupt the spread of fraudulent labor recruitment under Human Exploitation or Fraud, Scams, and Deceptive Practices policies across platforms and to provide additional protections to users, Meta should introduce an informative notice. It could be triggered when users are engaging with (react, comment, share or click on an external link) the content that is flagged by its technology as involving signals of job fraud and recruitment into labor exploitation but left on the platform due to low or medium levels of confidence for removal.
The Board will consider this recommendation implemented when Meta confirms that the new informative notices are provided to users in all languages supported by the platform.
Our commitment: We have introduced a range of product interventions to help disrupt fraudulent labor recruitment and reduce the likelihood that users are potentially impacted, including warnings and friction that help users pause and make more informed decisions when interacting with potential job scams. We consider this recommendation implemented and will maintain our commitment to ongoing improvement as scam tactics evolve.
Considerations: It is important to us that users can curate their experience across our platforms. Our algorithms on Facebook, Instagram, and Threads aim to help people connect with content and creators they find valuable, and we work to ensure this is balanced with our commitment to protect users from harm. To support that commitment, we deploy a range of interventions designed to intercept scams and exploitation and to reduce the likelihood that users will be victimized.
On Facebook, we currently deploy search-triggered interstitials across several high-risk markets to help prevent people from engaging with fraudulent job posts linked to criminal scam syndicates. These interstitials are designed to address content related to labor exploitation and forced criminality associated with scam centers. Functionally, the interstitial surfaces a warning message when users type keywords associated with high-risk fraudulent job advertisements that may lead to labor exploitation. It blocks users from viewing search results until they choose to proceed, and it connects potential victims to organizations that can provide immediate support and resources. Where feasible, the interstitials are translated to support local languages, and our Safety and Operations teams collaborate to compile and maintain a representative, regionally nuanced keyword set associated with this form of labor exploitation. We continuously monitor the performance and impact of these intercepts and assess how to update keywords to reflect evolving adversarial behavior.
On Messenger, we have invested in more advanced scam detection in chats. When enabled, and a new contact sends a message that appears potentially scammy, we surface a pop-up warning and give the user the option to report the message, which enables a review of the chat by AI. Where a scam is detected, we provide relevant recourse information and support resources. We’ve also recently begun integrating AI into our content moderation system to better address this type of violating content. For example, a common scammer tactic is to message advertisers, impersonating Meta to scare them into revealing their password so the scammer can steal the account. When we launched a new LLM, we detected and mitigated an additional 5,000 scam attempts per day that no existing review team had caught.
We also rely on transparency mechanisms to provide additional context on our interventions and to support our broader approach to user notices. Our Taiwan Fraud Prevention Transparency Report describes our fraud and scam strategy across four pillars: (1) build platform defenses, (2) disrupt threat actors, (3) enlist stakeholders to prevent and respond to fraud and scams, and (4) empower users. This reporting provides additional insights and relevant metrics illustrating how we work to disrupt fraudulent labor recruitment.
Recognizing the breadth and variability of scams related to fraudulent labor recruitment, we also maintain a scam prevention hub dedicated to empowering users through education and tools that help them protect themselves across our platforms, including guidance and education on how to identify and distinguish between different types of scams.
Our work to disrupt fraudulent labor recruitment and other fraud activity is dynamic and iterative. We continuously collaborate across teams and draw on external insights to ensure our approach remains responsive to changes in adversarial behavior. Given the systems already in place and our ongoing operational workflows, we consider this recommendation complete and will have no further updates.