On 21 October 2026, eSafety Commissioner Julie Inman Grant appeared in Canberra alongside ASIO, the Australian Federal Police and the Department of Home Affairs. The occasion was the first hearing of the Parliamentary Joint Committee on Intelligence and Security's inquiry into the nature and drivers of radicalisation in Australia (Cyber Daily). The inquiry matters beyond security policy. It will help decide how far Australia's online safety regime stretches into encrypted messaging and gaming services, and whether that stretch is justified by what the regime has already delivered.
What the inquiry is examining
The committee began the inquiry on 14 September 2026 after a referral from the Minister for Home Affairs. Its terms of reference cover pathways to radicalisation, the role of technology and encrypted communications, how algorithms may amplify extremist content, and the grooming of minors, including through gaming and gaming-adjacent platforms. They also cover the obligations of platform, gaming and messaging providers to detect, report and respond to extremist content. Submissions close at the end of Friday 30 October 2026 (Medianet).
The hearing came two days after an alleged attack on independent MP Helen Haines outside her Wangaratta electorate office on 19 October. Chair Senator Raff Ciccone called the first hearing a way to build an evidence base on a "complex and evolving challenge" (Cyber Daily).
The strongest case for tougher rules
The case for action is serious and should be stated fairly. AFP Assistant Commissioner Sandra Booth told the inquiry that AI is "accelerating and lowering the skills required to actively manipulate and radicalise victims online." She said joint counter-terrorism teams have investigated children as young as 13, including a boy who allegedly kept school-shooting plans in a notebook. ASIO's deputy director-general Claire McGuinness put the national terrorism threat level at "probable" and said the environment is becoming more permissive of violence (AAP via Port Macquarie News). If radicalisation that once took months now takes weeks, a regulator relying on after-the-fact complaints looks structurally behind.
What the existing powers have and haven't shown
The weakness in the argument for expansion is that the most ambitious recent tool is still being evaluated. The social media minimum age obligation, built on the Online Safety Act 2021 (legislation), took effect on 10 December 2025. Penalties reach A$49.5 million for companies. In March 2026 eSafety said it was investigating five platforms (Facebook, Instagram, Snapchat, TikTok and YouTube) after issuing 23 legally enforceable notices to 10 platforms, and it aimed to make enforcement decisions by mid-2026 (Clayton Utz). eSafety also reported 4.7 million under-16 accounts restricted, while acknowledging that a substantial number of children still hold accounts or get past age checks (eSafety).
Those figures show two things. Compliance machinery can move large numbers quickly. But the number of accounts removed is an input, not an outcome. Nothing in the material I could verify shows that removing accounts has reduced radicalisation, grooming or harm to minors. That is why the evaluation eSafety began in February 2026 matters more to this inquiry than any new mandate.
Where proportionality should bind
Three principles should guide the committee.
- Leave encryption intact. Duties to "detect" extremist content on end-to-end encrypted services can only be met by scanning content before it is encrypted or by weakening the encryption. Either approach exposes every user, including the journalists, dissidents and abuse survivors who rely on private channels. Targeted lawful access to specific suspects through existing warrant powers is a different and narrower thing.
- Regulate process, not speech. Most of what drives radicalisation is lawful, if ugly, speech. Obligations that require platforms to publish risk assessments, report on recommender-system design and respond to credible referrals are auditable and contestable. Takedown quotas for vaguely defined "extremist" content push platforms to over-remove, and the burden falls on political dissent and minority communities.
- Match gaming rules to evidence. eSafety has so far treated Roblox as primarily a gaming service outside the age rule. It has instead used transparency notices, with daily penalties of up to A$825,000, on Roblox, Minecraft, Fortnite and Steam (Clayton Utz). That is the right order of operations: collect data, then decide whether a mandate is warranted. Extending the age rule to games before those results are public would reverse it.
The stalled alternative
The political debate is also lopsided. The government has banned social media for under-16s, but its proposed digital duty of care, which would also let users opt out of algorithmic feeds, remains stalled. The Coalition calls it political censorship and the Greens want tougher penalties (AAP via Port Macquarie News). A systemic duty of care, judged on whether a platform's design and processes are reasonable, is a better fit for a problem rooted in recommender systems and private channels than another prohibition on a category of users. It still needs tight drafting so that "harm" is defined narrowly enough to protect lawful expression.
What to submit by 30 October
The committee has an unusually good opportunity. It can ask eSafety for outcome data from the minimum age regime, including whether displaced teenagers moved to less moderated services. It can ask the AFP and ASIO which online pathways, in the cases they have seen, a platform duty could actually have interrupted. And it can ask whether the harm lies on services already covered or on ones that would be newly pulled in. A regime that can show it reduced harm earns the case for more authority. One that cannot yet show this should say so before it asks for more.
I could not verify what Ms Inman Grant told the committee on 21 October, so this analysis rests on the inquiry's terms of reference, the other agencies' reported evidence and eSafety's published compliance record, not on her testimony.