Meta Director Denies Policy Changes Led to Rise in Antisemitic Content

A policy director at Meta has rejected claims controversial changes to the social media giant's content moderation policies led to an increase in antisemitic content across its platforms.

The company, which has 3.5 billion users across Facebook, Instagram and Threads, on Monday fronted the Royal Commission on Antisemitism and Social Cohesion, which is this week examining the prevalence of online hate.

Meta boss Mark Zuckerberg in January 2025 announced sweeping changes to its Community Standards on Hate Speech to "get rid of a bunch of restrictions" on how people could discuss topics like gender and immigration.

Mr Zuckerberg also announced changes to the way hateful conduct violations would be enforced in response to concerns about over-enforcement, and the removal of content that did not violate Meta's community standards.

"It means we're going to catch less bad stuff, but we'll also reduce the number of innocent people's posts and accounts that we accidentally take down," he said in a video announcement last year.

Meta core policy director Benjamin Good told the royal commission the changes meant enforcement of hateful conduct standards was now mostly reactive, relying on users to report harmful content to the platform.

He said this was in contrast to the "gold standard" proactive approach to content moderation which used artificial intelligence tools to remove harmful posts before they could be viewed.

Mr Good said these tools were largely used to manage illegal and high-severity violations such as terrorism, drugs, frauds and scams.

"The reactive approach differs from the proactive approach in that it is less prone to over-enforcement, and that was the problem motivating the changes," he said.

"It's true there may be content that we don't remove because it is not reported, but I do want to emphasise we are very carefully monitoring the extent to which that is the case."

Counsel assisting Richard Lancaster SC put to Mr Good the reliance on user-reported violations had resulted in "substantial" complaints from Jewish advocacy groups that more antisemitic content was being shared on Meta's platforms.

"I would not characterise the engagements [with Jewish advocacy groups] that way," he said.

"Those conversations I would describe as overwhelmingly productive and constructive.

"Of course if we miss content, we should learn from that and get better, and I have had conversations with groups whenever they see trends that they've identified that they think we should be removing."

Less hate speech content removed

Data presented to the inquiry showed the trend of content that was removed from Facebook and Instagram for violating hateful conduct standards before and after the policy changes were introduced.

About 5.8 million items were removed or otherwise acted on by Facebook for hate speech from October 2024 to December 2024.

By contrast, just 1.2 million items of hate speech were removed from July 2025 to September 2025, after the policy changes were in place.

On Instagram, Meta took action on 7.4 million items of hate speech from October 2024 to December 2024.

That figure dropped to 2 million in July 2025 to September 2025.

Royal Commissioner Virginia Bell pressed Mr Good on the figures.

"Can you suggest a plausible explanation for a drop in say 79 per cent in the metric of hateful conduct content other that the announced change in the policy that occurred in January 2025?" she put to Mr Good.

"I truly don't know and I don't want to speculate," he said.

"I don't know if this relates solely or entirely to a switch to reactive enforcement."

"[Reactive enforcement] still remains a very effective mechanism because when content gets views it tends to be reported if it is potentially violating and then removed."

The inquiry heard the royal commission alerted Meta to at least four antisemitic posts that appeared on Facebook during the hearings, with the content initially not deemed to have violated community standards.

It was subsequently removed after a review of the decision was sought.

Meta prepped staff responses​

Internal Meta documents shown to the inquiry outlined suggested responses and talking points for staff asked questions about community guideline changes.

When asked why phrases like "immigrants are scum" were not considered dehumanising under the new policy, the document suggested staff highlight it was important to distinguish between offensive speech and content that could lead to offline harm.

The document said Meta would continue to prohibit attacks based on protected characteristics such as ethnicity, race or religion.

Mr Good said the policy changes were not aimed at changing Meta's approach to antisemitism.

"They were aimed at making small changes to the policy that would allow people to say things on our platforms that they were saying in congress, on the news and the goal of that was because we want people to be able to debate these political topics," he said.

Commissioner Bell again pressed Mr Good on the effectiveness of Meta's policy shift in January last year.

Given the nature of the material … does it cause you to revise your opinion that user reporting tends to work well?" she asked.

"It's always hard to know why a particular post was missed and I don't know why that was the case in those four instances," he said.

"Each of those four instances I personally reviewed … and I can commit to learning from them and address those moving forward."

"But on a platform of our scale, billions of users, we have to be focused on metrics which show risk of potential harm and for me that is views of violating content."