The firm has been asked to change its cross-check programme by an independent oversight group that evaluates Meta's content moderation choices, and the company has sort of agreed.
The method placed content from "high-profile" users in a separate moderation queue from the automated one the firm uses for normies. The Oversight Board, the "independent body" that oversees Meta's content moderation choices, gave 32 proposals for changing the programme. Certain public figures including politicians, celebrities, and athletes have their flagged content remain up "waiting further human review" rather than being removed.
In direct response to a 2021 Wall Street Journal article that looked at the exempted, the Board conducted its study. Although "a content review system should treat all users similarly," the board noted the inherent difficulties of managing large volumes of content in their conclusion, noting that the programme has "broader challenges in moderating huge volumes of content."
For instance, they claim that Meta was undertaking so many daily moderation efforts at the time of the request — roughly 100 million — that even "99% accuracy would result in one million mistakes every day.
The Board notes that the cross-check programme was "more directly oriented to serve business concerns" and was less focused on "advanc[ing] Meta's human rights obligations."
The Board offered 32 changes to the cross-check programme, of which Meta decided to implement 11, partially implement 15, continue to consider one's viability, and forego the remaining five. The initiative will be "more transparent through frequent reporting," the firm stated in an amended blog post published on Friday. The program's eligibility requirements would also be adjusted to "better account for human rights concerns and equity." In order to lessen the backlog of review requests, the company will also modernise its operating systems, which will speed up the review and removal of dangerous content.
Although the revisions "may render Meta's approach to error avoidance more fair, believable, and genuine," the Board stated in its Twitter thread that "many areas of Meta's response haven't gone as far as we advised to produce a more transparent and equal system."
Follow Us