In her testimony to Senate Estimates on Wednesday night, the eSafety Commissioner, Julie Inman Grant, noted in passing that the Australian Classification Review Board had recently overturned two decisions originally sought by the regulator.
I want to focus on one of the decisions overturned by the Review Board, as it highlights better than anything I have seen in my roughly 15 years of dealing with online harms that the National Classification Scheme is totally misaligned with the reality of regulating harmful online content.
In 2025, the eSafety Commissioner sought classification from the Board of a clip circulating on the X platform. (eSafety can refer items to the Board to obtain a ‘formal’ classification, which is an administrative decision.) The content depicted the shocking and abhorrent murder of a young Ukrainian refugee, Iryna Zarutska, in August of that year. Iryna was stabbed to death while on a train in Charlotte, North Carolina, by a male passenger, unknown to her.
The framework underpinning the assessment of content by the eSafety Commissioner is the National Classification Scheme. This arrangement, which provides specific statutory criteria for classifying material and related guidelines, was developed decades ago to guide the Classification Board in its classification of commercial films and publications. The Scheme was later expanded to include computer games. Typically, online content such as images and video clips are assessed as ‘films’ under the Scheme, which (it should be remembered) was created and is maintained to guide classification of commercial releases appearing on cinema screens, and streaming services such as Netflix.
The Board classified the Zarutska footage as Refused Classification (RC). A Freedom of Information request has led to the Board publishing a redacted version of its decision, which nevertheless confirms that the Board had regard to a clause in the National Classification Code that justifies a Refused Classification decision where ‘films’:
depict, express or otherwise deal with matters of sex, drug misuse or addiction, crime, cruelty, violence or revolting or abhorrent phenomena in such a way that they offend against the standards of morality, decency and propriety generally accepted by reasonable adults to the extent that they should not be classified
The Board’s classification decision is relevant to the steps eSafety next took. To explain, the Online Safety Act 2021 (Cth) provides that ‘class 1’ material can be subject to a removal notice issued by the eSafety Commissioner. Class 1 material is any material that would be Refused Classification by the Classification Board, and includes child sexual abuse material, certain violent material, and material that incites the doing of a terrorist act.
When a service such as a social media service is issued a class 1 removal notice in relation to specific material, a 24-hour compliance timeframe begins ticking. Failure to remove class 1 material attracts civil penalties.
Armed with the Board’s classification decision, the eSafety Commissioner issued a class 1 removal notice to X specifying 23 instances on the platform where the video appeared. The removal notice confirmed that a classification decision of RC had been made by the Board, and directed takedown of the clip from each X.com URL within the statutory compliance period of 24 hours.
X sought review of the decision by making an application to the Classification Review Board, a statutory body with the power to reclassify a Board decision. In part, X argued that “the most appropriate classification is R18+ (at the highest) which gives effect to the Code’s principle that adults should be able to read, hear, see, and play what they want.”
In an extraordinary moment of moral hypocrisy, X also argued that the impact of the clip showing Iryna’s murder was mitigated on the basis that “the release of footage depicting the attack on Iryna Zarutska prompted widespread political discussion and public debate on a range of issues including violence against women, mental health, and policies of crime.”
Given the way X.com has been deliberately weaponised to scale and automate abuse, online sexual violence, harassment and intimidation against women, the attempt to hold itself out as a facilitator of much-needed public debate about violence against women is breath-taking in its audacity.
Meanwhile, the eSafety Commissioner correctly asserted that the clip “gratuitously depicts the event of a graphic public murder … depict[ing] a level of detail of the event which is not required for the purposes of facilitating public commentary.”
The reasons stated by eSafety were as plain as they were necessary: “the film depicts a real-life murder”.
Applying the convoluted and legalistic language of classification, the Review Board found for X, noting (despite the sheer brutality of its own description) that:
[T]he depiction does not have a very high degree of impact because it lacks detail. The Review Board agrees with the eSafety Commissioner that the depiction contains some detail, including the striking motion made by the attacker, the victim’s blood on the seat, the blood spatter on the floor, and the victim’s shocked expression as she collapses after the attack. However, the Review Board does not consider the depiction to be excessively detailed, noting factors that significantly limit or obscure detail, such as the brevity of the depiction, the blurring of the attack so that the weapon and stabbing impacts are obscured, and the distance of the camera from the victim – the vantage appears to be a fixed CCTV camera from a moderate distance.
Of the CCTV camera, the Review Board identified the fact the assault was captured by a CCTV system as relevant to its decision. A CCTV system “[is] a system that is widely understood to be a means of detecting public incidents and disturbances for the purpose of—among other things—identifying victims and offenders. The camera is a chance observer of the event, and not an active participant in the way that might be attributed to a film camera deliberately positioned to record a planned scene, for example.”
Why the Review Board laboured so awkwardly to draw a distinction between deliberate Hollywood framing of a scene, and the chance recording of a grisly crime by CCTV is one of the more baffling elements of its analysis.
Later in its decision, the Review Board turns its mind to the “standards of morality, decency, and propriety generally accepted by reasonable adults”. Its conclusion?
The Review Board accepts that reasonable adults will regard the violence depicted as abhorrent, and many will find the footage distressing. However, the level of detail in the depiction is not so graphic that it offends against community standards to the extent that it should not be classified. In the Review Board’s view, reasonable adults ought to be able to watch the film if they wish, with appropriate warning.
Let’s return to what this ‘film’ shows: the brutal, callous, and graphic murder of a 23 year-old refugee. Iryna dies quietly, bleeding out in front of a carriage filled with bystanders. A more horrific, lonely death is hard to imagine. Yet the Review Board saw it as their appropriate role to focus on narrow, technical criteria, instead of the cruel and inhuman extinguishing of a life. Their final judgement was that Iryna’s murder clip ought to be reclassified from RC to the lower rating of R18+.
Out of a dense thicket of specious and insensitive justifications for the lower classification, the Review Board ends with this extraordinary conclusion:
The film has been published elsewhere on the internet and is likely available to anyone who has access to the internet, including minors. The Review Board considers that, to refuse classification out of concern that minors may see a film classified R18+ would make the R18+ classification meaningless. Allowing this consideration to influence a classification decision is contrary to s 1(a) of the Code that adults should be able to read, hear, see and play what they want.
There are two serious issues with this. First, the fact that the material is available elsewhere on the internet is irrelevant to the question of classification. Worse, this statement implies that targeted regulatory action is futile when it comes to online harms. If that were true, then we might all now immediately cease our efforts to combat online child abuse content, or hate speech, or videos promoting the Islamic State.
Second, this isn’t just about minors seeing the clip. There is no remotely reasonable argument supporting the rights of anyone to view the murder of Iryna Zarutska. To assert that refusing classification in this case would infringe the Code’s foundational principle is a concession by the Review Board that, no matter how grotesque or gratuitous the content, it should be available for all to see. This argument only appeals to ghouls who use the thin screen of “free speech!” to legitimise their voyeuristic interest in graphic violence.
Videos like those showing Iryna’s murder are precisely the reason why the Online Safety Act was created. Government recognised that eSafety needed to have faster powers to order the removal of harmful content without having to procedurally rely on the Classification Board for classification, as was the case under the predecessor legislation. The aim was to create a rapid removal scheme for the worst of the worst content, such as videos depicting violent murders, child sexual abuse material, and pro-terror media.
But Government flinched at taking the step needed to make the Act truly effective, which would have involved de-coupling Part 9 from the National Classification Scheme.
In her review of the Online Safety Act, Delia Rickard PSM in 2024 recommended this very step. Rickard noted that the Scheme “is not workable as the basis for a regulatory regime designed to apply to vast volumes of online content, that as eSafety suggest in their submission, is dynamic, fluid and even ephemeral.”
Her recommendation was to establish “clear rules to determine whether certain material is illegal or harmful in order to trigger rapid removal, appropriate regulatory action and efficient compliance by online services.” It is a position that won broad support from across industry, including Meta, Telstra and the Digital Industry Group Inc, all of which argued for clearer definitions than the convoluted and subjective construction of the Classification Code.
Rickard recommended creating a taxonomy of harms that would be defined in the primary legislation. This taxonomy would include ‘abhorrent violent conduct’ (AVC) material, taking its definition from the Criminal Code Act. Under section 474.32 of that Act, AVC material includes material depicting a person engaging in acts such as murder, attempted murder, terrorism, torture, rape or kidnapping.
Such a taxonomy would make the work of assessing and declaring content class 1 material far more straightforward. Media depicting child abuse material or violent crime shouldn’t require an assessment of the general moral orientation of the reasonable adult, nor the content’s artistic or educational merit. There is no merit to material valued only for its capacity to shock or arouse the worst interests of the worst people online.
In April, the new Minister for Communications, Anika Wells, released the Government’s response to Rickard’s 67 recommendations for reform of the Act. Two of these advised removing the link between the Act and the Classification Scheme, and creating a new set of harmful content definitions and thresholds. The Government’s response was merely to ‘note’ the recommendations. This means they have gone into a holding pattern, perhaps to be revisited at some point, but likely not in this parliamentary term.
This is a missed opportunity to reforge the Act to be responsive to the way online content harms are actually perpetuated. Whether an antiquated and anachronistic classification framework (the classification guidelines were last updated in 2012) was ever fit for the purpose of addressing high harm internet-based material is debatable. But the misalignment today is so great and so glaring, so divorced from public sentiment, and so antithetical to the Online Safety Act’s legislative intent, that it has become a colossal roadblock frustrating effective action against anything other than child abuse material.
In its decision to reclassify Iryna’s murder video, the Review Board has created an indefensible, appalling precedent. It is one that will undoubtedly now be used by X to argue against the merits of any attempt by the regulator to hold the platform to account for its abandonment of any semblance of moral leadership. The line needs to be drawn somewhere, and in a way that genuinely reflects community standards.
Clearly, the Classification Scheme cannot achieve this goal.