Facebook and Instagram parent Meta faced tough questions from Australian officials investigating antisemitism and hate speech online. Benjamin Good, who runs Meta's policy decisions worldwide, testified on July 6, 2026, about why his company changed how it polices harmful content—and whether that change let hateful material spread unchecked.

Why This Hearing Matters

Australia's government launched a formal investigation into antisemitism and online hate speech after a terrorist attack in Sydney in December 2025. The investigation, called a royal commission, held its first public hearing in May 2026 and has been digging into whether social media platforms are doing enough to stop hateful posts. The July 6 session marked the first time officials directly questioned Meta about its policies.

What Changed at Meta in 2025

In January 2025, Meta announced a major shift. The company said it would stop automatically catching and removing problematic content before users could see it. Instead, it would only remove posts that break laws or violate what it calls "very serious violations." The company also stopped using independent fact-checkers.

Meta's CEO Mark Zuckerberg explained the reasoning: if the company removes less bad content, it will also remove fewer innocent posts by mistake—posts that shouldn't have been taken down at all. The Guardian reported his statement.

But since January 2025, the royal commission received complaints that hateful posts, including antisemitic ones, were showing up more often on Facebook and Instagram. The counsel questioning Good also pointed out something concrete: people who had testified to the royal commission were later targeted with hateful content on Facebook.

What Meta Actually Allows on Its Platforms

During the hearing, officials reviewed internal documents that showed exactly what Meta's rules permit. Some examples from these documents:

• Posts saying "gay people are sinners" are allowed.
• Posts claiming "immigrants are criminals" or "white people are all Nazis" are allowed.
• A post saying "Black people are more violent than whites" is allowed, but one saying "black people are all drug dealers" is not.

The difference, according to Meta's own rules, comes down to how specific the claim is. One applies a trait broadly; the other pins a specific crime to a group.

Meta's internal document states plainly: "It is not Meta's role to police offensiveness."

This distinction—between something offensive and something that directly calls for harm—sits at the center of how Meta operates now.

Meta's Defence

Good did not argue the documents were wrong. Instead, he pointed to a different problem: aggressive removal of hateful content backfires. He described the ideal approach as catching all hateful material before anyone sees it, but he said that goal creates real damage. Jewish users, he noted, had their own posts removed when they spoke out against violence during crises. Posts opposing Hamas, which the Australian government lists as a terrorist group, were caught in the same net and deleted by mistake. These, he called them "very bad outcomes."

Good also told the commission that Meta's own measurements show the amount of hateful conduct violations has stayed flat at 0.02% of all content since 2022—even after the January 2025 change. He offered this as proof the system was working.

There is reason to be cautious about that statistic. When a company measures violations, it can only count what its detection systems are built to catch. If the company removes fewer automatic safeguards, more harmful content may simply never get flagged in the first place, making the violation rate look steady even if harmful posts are actually increasing.

The Bigger Picture

This hearing was one of several dense technical sessions at the royal commission. Just days earlier, experts told the commission that far-right groups are deliberately using hidden symbols and memes designed to slip past automated content filters, according to ABC News. The commission also learned that Australia is working on new laws to hold social media platforms accountable—similar to how the government regulates television—but those laws are still years away from being ready, per ABC News.

Beneath all the testimony is a structural reality: Meta sets its own rules and enforces them itself, with little outside government oversight at the moment. Good's evidence, whatever weight it carries, reflects that basic truth—Meta decides what is acceptable on its own platforms.

The royal commission's job is to investigate whether this arrangement is still adequate. The political moment that created the investigation suggests Australian policymakers are losing confidence in it. Whether the commission's final recommendations will have legal force—and what those recommendations will be—remains unclear. The remaining hearings will shape that answer.