Complaints to Australia’s online safety regulator about terrorist and violent extremist content nearly quadrupled last financial year, a parliamentary inquiry has heard.
Heidi Snell, general manager of regulatory operations at the eSafety Commissioner, told the Parliamentary Joint Committee on Intelligence and Security that the agency received more than 2,100 such complaints in 2025-26, up from about 550 the year before.
The commissioner’s written submission (pdf) says the jump may reflect community concern and reporting habits rather than a direct measure of how much of the material is out there.
Snell took on notice questions from committee chair, Senator Raff Ciccone, about how many of those complaints led to content being taken down.
Kathryn King, general manager of the commissioner’s corporate and strategy division, said more than half of young Australians had come across this kind of material by accident, often through the automated feeds and algorithms that dictate social media feeds.
“Young people are not seeking out this type of material online proactively,” she said.
King told the committee these feeds favour whatever draws attention rather than weighing what the content is, adding, “We hate to say it, but of course, there’s more engagement often on harmful material.”
AI Safety Measures Stripped Away
Mike Skwarek, head of codes and standards enforcement, pointed to testing by the group Tech Against Terrorism, which found that 17 percent of 116 AI models would give instructions on how to carry out a terrorist attack.
He said the safety features built into openly released models were often removed within three days, and that one Meta model, which scored 97 out of 100 on the group’s safety test, scored just three once those features were stripped out.
He said the model-sharing site Hugging Face hosted about 29,000 of these altered versions, but had since changed its rules after talks with eSafety.
The regulator began civil penalty proceedings against Telegram in the Federal Court in July, alleging the messaging service failed to detect and remove pro-terror material, including videos of terrorist executions and mass shootings.
Issues With Social Media Ban
Deputy Chair Andrew Wallace told the hearing that students at a school he visited said one boy whose account was shut down under the under-16 social media ban opened a new one the next day.
Snell said platforms had reported deactivating or removing about 4.7 million accounts and that investigations into five major platforms were continuing, but stressed that “individuals’ experiences are not evidence of a failure to take reasonable steps.”
King said the regulator’s own two-year study had found that more than 50 percent of young people had not even been asked to check their age.
Classification Regulation Not Fit for Purpose
The eSafety’s submission says the Classification Board rated footage of three killings in the United States, including the assassination of Charlie Kirk, as “refused classification,” meaning it cannot legally be shown in Australia.
But after X Corp appealed, some copies were given lower ratings by the Classification Review Board, leaving the regulator unable to order their removal.
Snell said the classification system was “a challenge for us in being able to be immediately responsive.”
Platforms almost always challenged removal notices issued before a formal rating, she said, and were now challenging formal ratings too. She added that the scheme was built for commercially made films, not videos uploaded by ordinary users who have little idea of the rules.
General Counsel Richard Fleming said the regulator’s main tools were court-based, and when a company deliberately wanted to show content to Australians, “we don’t have a lot of levers to pull.”
English (US) · 
Turkish (TR) · 