ABC News Afternoons 7 July 2026

Hearings to probe how YouTube, TikTok manage risks of online hate

ABC News Afternoons

Transcript from interview with Gavin Coote.

Yeah, pretty much all the major social media platforms are using AI, and some of the deficiencies in AI have really been a point of focus in the Royal Commission.

But today, we've actually learned of an example of where AI actually overruled an error made by human moderators. So, a piece of content came to the attention of the Dor Foundation, which is a group tackling anti-semitism co-founded by the former Federal Treasurer, Josh Frydenberg.

It found a piece of online anti-semitic content on TikTok in May. That actually, initially, auto the AI moderators had found that content to be anti-semitic in violation of the TikTok guidelines and was sort of effectively shadow-banned.

It wasn't allowed to be published. Then the creator got a review. He asked for an appeal. Human moderators, two human moderators, looked at it and said, look, this is okay.

We're actually going to allow this to be published. It remained online for about a month until the Dorr Foundation escalated that issue. And they found that the human moderators had made an error in that.

They allowed this to go up, despite it violating the community safety principles. So, here is Zachary Hecht. He's the Global Head of Policy with Trust and Safety with TikTok, explaining what has happened since then, that the post has been removed, and that he feels that there might need to be some fine-tuning around this policy.

Our focus is on getting content moderation right at scale for billions of pieces of content.

This program is not uploaded to ABC iView