Meta executives faced intense questioning at Australia's royal commission into antisemitism and social cohesion on Monday, as evidence emerged that the company's decision to reduce content moderation may have unleashed a surge in hateful conduct on its platforms. The inquiry was shown graphs indicating a steep 79% drop in enforcement actions against hateful content since January 2025, when Meta announced it would 'reduce censorship' and rely more on user reporting.
Policy Shift and Its Consequences
In January 2025, following Donald Trump's re-election, Meta CEO Mark Zuckerberg announced the company would eliminate factcheckers and only proactively tackle illegal and very serious violations. Zuckerberg described it as a 'trade-off', acknowledging it would catch less harmful content but reduce accidental takedowns of innocent posts. Benjamin Good, Meta's global director of core policy, told the commission that the changes aimed to avoid 'over-enforcement', which he said posed 'significant risk to the communities that we try to protect'.
However, counsel assisting Richard Lancaster argued it was 'entirely unrealistic' to think the policy shift hadn't altered how moderators operated. He pointed to witnesses being targeted on Facebook after giving evidence and internal documents showing offensive comments like 'gay people are sinners', 'immigrants are criminals', and 'black people are more violent than whites' are now permitted, as long as they don't charge specific criminal behaviour.
Internal Guidance and Enforcement Data
The inquiry saw Meta's internal 'frequently asked questions' guide, which states: 'It is not Meta's role to police offensiveness.' According to the document, false claims such as 'white people are all Nazis' are allowed, but 'black people are all drug dealers' is not, because it implies specific criminal activity. Good defended the approach, saying the 'gold standard' is removing hateful content before it's seen, but that over-enforcement could censor legitimate speech, such as Jewish communities speaking out against atrocities.
Lancaster challenged Meta's claim that the prevalence of hateful conduct violations remained at 0.02% since 2022, arguing that 0.02% of hundreds of billions of pieces of content is still a large absolute number. Commissioner Virginia Bell pressed for a 'plausible explanation' for the 79% drop in enforcement actions, other than the policy change. Good replied that the ecosystem was complex and he couldn't attribute the drop solely to policy changes.
Expanded Hate Speech Policies
Despite the moderation cuts, Good highlighted that Meta had expanded its hate speech policies to ban the use of 'Zionist' as a coded term to attack Jewish people. He explained that after consulting counterterrorism and antisemitism experts, Meta determined that many users were employing 'Zionist' to evade enforcement against claims that Jewish people have undue control over media or government. 'So instead of saying 'Jewish people do', they would say 'Zionists do',' Good said, adding that such claims are now prohibited.
Facebook Australia's public policy director, Mia Garlick, also appeared before the inquiry. She confirmed that Australia still retains factcheckers for misinformation, unlike other regions where 'community notes' have replaced them, but said the feature may be rolled out locally.
