The Drop a Royal Commission Asked Meta to Explain
In early July 2026, Australia's Royal Commission on Antisemitism and Social Cohesion pressed Meta and TikTok over how they police hateful content. Figures put to the inquiry showed the amount of hateful material Meta acted on fell by about 79 percent after the company loosened its moderation rules.
Meta and TikTok sent senior policy executives before Australia's Royal Commission on Antisemitism and Social Cohesion at public hearings in Sydney on 6 and 7 July 2026. The commission, established to examine antisemitism and social cohesion in Australia, used the sessions to scrutinize how the platforms detect and remove hate speech, and how their recent policy choices have changed what stays online.
The central figure was a decline. Data presented to the inquiry showed that action Meta took on hateful conduct on Facebook fell from about 5.8 million items in the last quarter of 2024 to roughly 1.2 million items between July and September 2025 — a fall of about 79 percent — according to Information Age, the publication of the Australian Computer Society. On Instagram, the reported figures fell from about 7.4 million items to 2 million over the same periods.
A number the commissioner wanted explained
Royal Commissioner Virginia Bell asked Meta's global director of core policy, Benjamin Good, to account for the 79 percent fall in the hateful content the company had acted on since changing its approach, Radio New Zealand reported. Good agreed the number was roughly correct but said he did not know whether it related “solely or entirely to a switch to reactive enforcement,” and declined to speculate further.
Meta changed how it moderates content at the start of 2025, moving away from systems that proactively scan for and remove certain categories of hateful material and toward a model that relies more heavily on users to report it. Good described the company's earlier, more proactive approach as the “gold standard” but said it also carried “a significant risk of over-enforcement,” according to Information Age.
Meta has rejected the suggestion that the shift produced more antisemitic content on its platforms. Good told the commission the company's engagements with Jewish advocacy groups had been productive, even as counsel assisting the inquiry pointed to complaints from Jewish organizations that such material had become more visible, Startup Daily reported.
TikTok's figures, and a comment reinstated
TikTok's global head of policy, trust and safety, Zachary Hecht, told the commission the company tends to remove potentially harmful content proactively in order to err on the side of caution. TikTok removed more than 270,000 videos that breached its safety and civility guidelines out of some 336 million posted to the platform in the 2025 calendar year, and more than 67,000 of about 110 million videos in the first quarter of 2026, Information Age reported. The company said 98 percent of that content was removed proactively in Australia in the first quarter, before any user report.
Hecht said accounts that incite violence against Jewish people would be “banned right away,” potentially after a single instance, Startup Daily reported. He also described a case in which an antisemitic comment correctly blocked by TikTok's automated systems was reinstated by a human moderator on appeal, and was taken down again only after a community organization reported it and the matter was escalated.
A recurring theme was that the platforms cannot readily quantify antisemitism itself. Removals are logged against broad rule categories such as hate speech or hateful ideologies, not by the group a piece of content targets, meaning the companies could not tell the commission how much of the material they act on is specifically anti-Jewish, The Canberra Times reported.
Why the numbers matter
Researchers who study platform moderation note that enforcement figures are among the few external measures of how much harmful content a company is catching. A large fall in the volume of material acted on can reflect either a genuine decline in such content or a decision to look for less of it. Meta's 2025 move from proactive detection toward user reporting shifts more of the burden of finding hate speech onto the people it targets, as the commission heard.
Public inquiries have become one of the principal ways governments extract such data from platforms that otherwise disclose it selectively. The Australian commission's hearings put the companies' own numbers on the record and required their executives to explain them in person, part of a wider pattern in which courts, regulators and inquiries press for accountability that the platforms have been slow to provide. The commission is due to continue taking evidence before reporting its findings.