Este contenido aún no está disponible en Español (España)
Alleged Audio Call to Rig Elections in Iraqi Kurdistan
ÚLTIMA ACTUALIZACIÓN 22 AGO. 2025
2025-019-FB-UA
Today, February 27, 2025, the Oversight Board selected a case appealed by a Facebook user regarding an audio clip posted by a prominent media outlet in Iraqi Kurdistan ahead of their parliamentary elections. The clip, overlaid with images of Patriotic Union of Kurdistan (PUK) leaders Bafel Talabani and Qubad Talabani, appears to depict a conversation between them discussing election rigging. The accompanying caption states the clip is a “recorded conversation” about “their sinister plans” for the Kurdistan region. According to reporting, the media outlet that posted the clip is affiliated with the Kurdistan Democratic Party (KDP) – the other main political party in the region. Reviews of the audio suggest it was digitally created or altered.
Meta determined the post did not violate our Misinformation policy, according to our Facebook Community Standards, and left the content up.
Under our Misinformation policy, we remove content that is likely to directly contribute to interference with the functioning of political processes. However, we may also place an informative label on realistic sounding, digitally created or altered audio that generates a “particularly high risk of materially deceiving the public on a matter of public importance.” We added labels to some posts containing the audio clip, noting the content may be digitally created or altered to seem real, but not all posts in order to avoid the risk of false positives.
We will implement the Board’s decision once it has finished deliberating, and 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, June 19, 2025, on this case. The Board overturned Meta's decision not to label all instances of the content on Facebook.
Meta will comply with the Board's decision within 7 days.
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 as this case. For more information, please see our Newsroom post about how we implement the Board's decisions.
After conducting a review of the recommendations provided by the Board, we will update this post with initial responses to those recommendations.
Recommendations
Recommendation 1 (Implementing in Part)
To ensure “High Risk” labels are applied consistently to all identical or similarly manipulated content, Meta should apply the relevant label to all content with the same manipulated media on its platforms, including all posts containing the manipulated audio in this case.
The Board will consider this recommendation implemented when Meta provides the Board with a clear process for consistently identifying and applying the appropriate “High Risk” label to all instances of manipulated media across the platform.
Commitment Statement: We recently updated our internal guidance in line with aspects of this recommendation, and we now consider banking content when we apply a high-risk label to content. This process allows for further application of the label to identical content that is not shared in order to condemn or debunk.
Considerations: In alignment with a previous recommendation from the Oversight Board and input from expert stakeholders, we apply high-risk labels to content that may materially deceive the public on a matter of importance. This more prominent label may be applied in addition to other AI info labels that we apply to the content to provide context. We recently updated our internal guidance so that, when we determine that the high-risk label is necessary, we also apply the label when we identify duplicate, identical content. However, we also recognize that the identical content may be shared in a clearly condemning or debunking context and currently would not apply the label in those cases. For example, if a user reshared an AI-generated video of a world leader saying something he or she did not say, but the user explicitly stated “this video is clearly AI and false,” we would not necessarily apply the label in that instance, as the explicit confirmation in the caption that the video is AI-generated means that content is considerably less likely to deceive the public.
Given the recent updates to our internal guidance, we consider this recommendation complete. Recommendation 2 (Implementing in Part)
As part of its electoral integrity efforts and to ensure users are informed of manipulated media on Meta’s platforms in the lead-up to an election, Meta should ensure that the informative labels for manipulated media on Facebook, Instagram and Threads are displayed in the same language that the user has elected for its platform. The Board will consider this recommendation implemented when Meta provides information in its Transparency Center about the languages in which manipulated media labels are available to users on its platforms.
Commitment Statement: Currently, we support informative labels, such as the “high-risk” label, in over 70 languages- covering a substantial majority of the languages available on our core services. We recognize that our billions of users speak a diverse range of languages, and we are committed to further improving parity between the languages used in our apps and those available for informative labels.
Considerations: We currently provide informative labels in multiple languages, recognizing the diversity of our global user base. We are continuously enhancing our systems to standardize our users’ language experiences to the extent feasible.
Our current system supports global accessibility by allowing internal teams to request translation support for our products and features, including informative labels, through a process managed by designated localization managers. These localization managers assign requests to specialists familiar with that specific product, feature or area. Through this system, the translation requests are prioritized based on various considerations. These teams work collaboratively to gather localizability feedback on the requests to understand if the terms are likely to translate smoothly into different languages. Additionally, this process provides guidelines and protocols for translation requests for features such as prioritization and turnaround times.
Although we currently translate many features, including labels, we do not translate labels in all supported languages. Our existing system is designed to manage and prioritize translation requests to ensure that the application of relevant labels remains consistent across our platforms and linguistically appropriate to reflect nuance and dialect. We will evaluate how to further refine this. We will explore collaboration across teams to ensure that there is increased parity between our supported languages—elected by users across our platforms—and available translations of our informative labels. We will provide updates on this effort in future reports to the Oversight Board.