A Law Unto Itself

Not content with the extensive suite of powers it already has to censor the online speech of Australians, the Office of the eSafety Commissioner has developed informal measures to achieve its ideological ends.

In the maze of government offices across Canberra, thousands of civil servants go about their work in anonymity. These bureaucrats supposedly implement policies developed by elected officials, feeding information up and down hierarchies. They are bound by departmental protocols and answerable to departmental secretaries, operating within defined limits of authority.

They are Australia’s public servants: an ever-growing class of unelected officials who remain constant across election cycles. Ordinarily, their work does not attract international controversy.

But in November 2025, something extraordinary happened. An Australian bureaucrat was summoned before a United States House Judiciary Committee to answer for her role in restricting Americans’ free speech; specifically, to defend actions taken under her initiative to force US company X (formerly Twitter) to take down footage of a church stabbing in the Sydney suburb of Wakeley.

Bureaucracies do not shrink.

The official at the centre of this controversy was Ms Julie Inman Grant, Australia’s eSafety Commissioner. A dual citizen, the accusation of her native country’s Congressman Jim Jordan was that Ms Inman Grant had become a “noted zealot for global takedowns” and was complicit in a “foreign censorship regime”.

Ms Inman Grant did not attend the hearing, risking possible contempt charges being laid against her. She later described the summons as “highly unusual”, noting she was simply “executing the will and the laws of Australia as passed by the parliament”.

But the truth is eSafety has become a power unto itself, with Ms Inman Grant’s nine-year tenure characterised by aggressive litigation designed to test and expand the limits of her office’s legal authority.

That Ms Inman Grant sees her role as testing Australia’s regulatory framework was evident in her comments before a panel at the Council of Foreign Relations, where she remarked, “When we were formed in 2015, we were the only online safety regulator in the world … we had to write the playbook as we went along.” By “writing the playbook” Australia’s eSafety Commissioner has become globally infamous as an internet censor-in-chief, the “noted zealot” of Congressman Jordan’s description.

MISSION CREEP

Significantly, when the role of the eSafety Commissioner was created, it was not given carte blanche to regulate the online activity of all Australians. Under the original Enhancing Online Safety for Children Act 2015 (Cth), the Commissioner was tasked with administering a relatively narrow complaints-based system for child cyber bullying.

At the time, the Institute of Public Affairs warned that the creation of such a position would lead to the censorship of online content for adults. Bureaucracies do not shrink. It is in their very nature to grow, employing ever-more staff to implement ever-more regulation. And how grave such a reality is when the regulator in question has the potential power to remove speech online.

In its 2014 submission to a comm-unications department consultation, the IPA warned against this inevitable “mission creep” and the extralegal growth potential of the agency:

There is no reason to believe that the Children’s e-Safety Commissioner will operate within the boundaries set by this parliament, nor any guarantee that its functions will not be extended in the future. Considering the substantial threat to freedom of speech represented by its proposed functions, the government should be worried that it is not creating a body that will become even more dangerous in the future. The only way to guarantee the Commissioner will not threaten free speech in years to come is to decline to establish it in this parliament.

As it turns out, the expansion of the role of Children’s eSafety Commissioner was not accidental, but a fulfilment of original intention. In December 2024, Ms Inman Grant explained that when Malcolm Turnbull, then Minister for Communications, introduced the original legislation, he “knew that we really had to start with children first, because you can’t really argue that children aren’t vulnerable”. In other words, children were not the endpoint of the regulatory agenda, but the starting point.

The scheme has grown to increasingly infantilise the entire adult population, evolving from a scheme to combat child cyber bullying to an expansive umpire intervening in the speech of Australian adults. This has occurred while a web of laws pertaining to speech remain fully operational, including state and federal anti-vilification laws, harassment laws, and incitement laws.

The Office of the eSafety Commissioner has also gained powers to remove content that is “refused classification” in Australia, namely violent content such as the livestreamed footage of the attempted murder of Bishop Mar Mari Emmanuel during a church service in Sydney in April 2024. Most recently, eSafety became the administrator of the under-16s social media ban, which came into effect at the end of 2025.

In numbers, its expansion has been meteoric. During its first year, the agency employed 37 people, with the Children’s eSafety Commissioner position receiving a base salary of $247,810 in 2016.

At the end of the 2025 financial year, the number of staff employed by eSafety had increased six-fold to 222 people, with the eSafety Commissioner’s base salary having increased to $463,781 (significantly above general inflation levels).

THE PROPENSITY TO LITIGATE: THE WAKELEY INCIDENT

Ms Inman Grant was appointed by the Turnbull government as eSafety Commissioner in 2017, following a career at Microsoft, Twitter, and Adobe. No other Australian government official has engaged in as many highly publicised and controversial lawsuits with big tech companies and individuals, stemming from their activities online, as Ms Inman Grant.

It seems the damage to Australia’s international standing is of no concern to the eSafety office.

This litigious approach reached its zenith following the terrorist attack on Bishop Mar Mari Emmanuel during the livestreamed church service in Wakeley, Sydney. When footage of the incident circulated online, the eSafety Commissioner issued takedown orders to Meta (parent company of Facebook and Instagram) and X, demanding removal within 24 hours under threat of significant financial penalties. Under the Online Safety Act 2021 (Cth) (the OSA) digital platforms are required to take “reasonable steps” to comply with formal takedown notices issued by eSafety.

While Meta promptly complied, X responded by geo-blocking the video content, meaning that users in Australia could not access the footage, except through use of a virtual private network (VPN), which encrypts a user’s internet connection, allowing them to mask their location and preserve anonymity.

This became the eSafety Commissioner’s justification to pursue X, demanding it remove from its global platform every possible way of viewing 65 links to the footage.

X rejected this as gross overreach. X’s head, Elon Musk, lambasted the “Australian censorship commissar”, asking, not unreasonably: if the global content ban was upheld “what is to stop any country from controlling the entire Internet?”

While eSafety initially succeeded in securing a temporary injunction, the Federal Court ultimately refused to extend it, with Justice Geoffrey Kennett ruling that a global ban of the footage was not “reasonable”. He concluded:

The argument that making the 65 URLs inaccessible to all users of X Corp’s platform everywhere in the world is not a step that it is ‘reasonable’ to require X Corp to perform in order to ensure that the URLs are inaccessible to Australian users (and therefore is not a step required by the removal notice) is powerful.

If given the reach contended for by the Commissioner, the removal notice would govern (and subject to punitive consequences under Australian law) the activities of a foreign corporation in the United States (where X Corp’s corporate decision-making occurs) and every country where its servers are located.

Somewhat prophetically, Justice Kennett went on to say that, had Ms Inman Grant’s edict against X over the Wakeley affair been upheld, it would have endangered amicable relations between nations, noting:

The potential consequences for orderly and amicable relations between nations, if a notice with the breadth contended for were enforced, are obvious. Most likely, the notice would be ignored or disparaged in other countries.

Justice Kennett’s concerns have materialised, with the office’s conduct having caused friction with Australia’s most important ally: the United States. Congressman Jordan specifically targeted Ms Inman Grant for her “expansive interpretation and enforcement” of the OSA, highlighting how her unilateral approach to regulation has drawn international attention. The rebuke has also damaged Australia’s reputation as a defender of democracy and shared Western values. In his letter, Jordan wrote:

Your expansive interpretation and enforcement of Australia’s OSA—including your claim of extraterritorial jurisdiction to censor speech outside of Australia—directly threatens American speech. Global content takedown orders are concerning because they harm the free speech rights of those outside of Australia’s jurisdiction and set the precedent that other governments may do the same.

For example, in 2024, your Commission sought to compel X to remove content globally, arguing that its geo-blocking of the content was insufficient because Australians could use VPNs to access the content. Other censorship regimes, like the one in Brazil, have used similar justifications when ordering global takedowns of content and threatening fines for VPN use.

The effect is that Australia is now grouped alongside authoritarian regimes in Washington policy conversations. A congressional report detailing Brazil’s censorship practices explicitly cited Australia as a “foreign censor” asserting extraterritorial jurisdiction and seeking to “impose a global censorship regime by ordering worldwide content removals”. Our reputation has thus been established, not for extending the virtues of democracy but for authoritarianism.

But it seems the damage to Australia’s international standing is of no concern to the eSafety office. In a statement announcing her withdrawal of the proceedings against X, Ms Inman Grant was unapologetic, claiming she had welcomed the opportunity to test her office’s legal powers:

Through this process, eSafety has also welcomed the opportunity to test its novel regulatory powers—set out under Australia’s Online Safety Act—to protect Australians from

online harm. We also welcome the Government’s Online Safety Act review currently open for public submissions, and the recently announced select committee inquiry into social media.

In other words, the Wakeley incident provided an opportunity for eSafety to test the limits of its powers in time for the government’s review of the legislation. In eSafety’s submission to the review, it called for increased “flexibility”, regarding the review as an “opportunity to hone the OSA’s strengths” and “ensure that changes are woven into an overarching logic that promotes effective action”.

INNOVATING TYRANNY: THE INFORMAL REQUEST REGIME

While the Wakeley case revealed eSafety’s willingness to exceed its authority, in truth a systematic pattern of regulatory overreach has operated in the shadows for years through the practice of “informal alerts”.

For years, eSafety has operated a sly system of content removal. Following a complaint of “adult cyber abuse”, eSafety informs platforms of the complaint and suggests they remove the speech as it violates the platform’s own terms of service. This approach effectively pressures platforms to remove content, conveniently obviating user appeal rights and shielding eSafety from accountability. It is an informal but highly effective means of censorship.

Fortunately, this practice was exposed in February 2026 through legal action brought by Sydney-based activist Celine Baumgarten. Baumgarten appealed the Australian removal of an X post where she protested the formation of a queer club for children as young as 8 in a Victorian school. Baumgarten had argued that the children were far too young to be exposed to gender ideology, criticising a teacher for their role in establishing the club.

While her post did not meet the threshold for a formal takedown notice, as there was no “intention to cause serious harm” to a particular Australian adult, as required under section 7 of the OSA, eSafety sought to have the post removed in Australia anyway by sending a complaint alert to X, the platform on which Ms Baumgarten’s message was posted. In doing so, it misrepresented its legal authority.

Specifically, eSafety staff acting on behalf of the Commissioner lodged the request through X’s high-priority legal challenge portal, crafting it to resemble a mandatory takedown order, and alluding to the OSA’s section 7 provisions on adult cyber abuse. This action relied on the appearance of legal authority to coerce compliance even while it was fully known that no such authority existed.

Baumgarten’s subsequent legal challenge in the Administrative Review Tribunal exposed the calculated nature of eSafety’s ploy. eSafety attempted to have the case thrown out on the grounds that the action was informal, and therefore outside the jurisdiction of the tribunal. This argument was emphatically rejected by the tribunal, which discerned eSafety’s attempt to have it both ways: wielding legal authority while claiming immunity from legal oversight. The tribunal concluded that eSafety could not create the impression of exercising legal power, achieve a coercive result, and then seek immunity from judicial scrutiny by arguing it acted outside its formal powers. When eSafety appealed the decision to the Federal Court, the decision of the tribunal was upheld.

The Federal Court judgment exposed the deliberate nature of eSafety’s regulatory abuse. As soon as it was established that the tribunal had jurisdiction over the matter, eSafety accepted it as inevitable that the tribunal would require it to reconsider its action. As Chief Justice Debra Mortimer and Justice Jonathan Beach wrote in their decision: “that is because it was obvious, and accepted by the Commissioner, that the statutory preconditions for the giving of a removal notice had not been satisfied”. In other words, eSafety continued to achieve coercive outcomes outside the legal process for as long as it could escape scrutiny, demonstrating a calculated disregard for the rule of law.

An exchange between staff from eSafety and the tribunal, cited in the Federal Court judgment, demonstrates how the Commissioner’s office has operated as a law unto itself to achieve the removal of what it considers objectionable speech:

Is it correct that eSafety’s policy is that taking informal or less intrusive action is preferred if it achieves the Commissioner’s desired regulatory results? — Yes.

In the context of this case, the desired regulatory result was that the applicant’s post would be removed from Twitter and Instagram? — Yes.

And that was the desired result, even though the investigator had concluded that the material in question was not cyber abuse material targeting an Australian adult? — Yes.

But in essence … in this particular case, the Commissioner through [her] staff concluded that [she] could not legally force the removal, correct me if I misunderstood your evidence, but the staff wanted to bring about the removal and wanted to achieve that result through the informal alert process, is that correct?

Through the informal alert process. Yes? — Yes.

The implications of this decision are far-reaching. The Court has established that the entire informal alert system—eSafety’s main tool for content removal—operates within reviewable legal territory, meaning the 1,036 instances of “adult cyber abuse” removed informally can be challenged in a tribunal. Yet it appears that the public servant responsible for this state of affairs will not be held to account.

And the scale of this shadow censorship regime is staggering: in the 2025 financial year alone, the eSafety Commissioner received 3,642 complaints of adult cyber abuse but only issued four takedown notices. Instead, the office initiated informal action on 219 occasions, and achieved a removal outside the formal process in 79 per cent of cases.

POLITICALLY DRIVEN CENSORSHIP DISGUISED AS DUTY OF CARE

It is one thing that the agency has acted extralegally and sought avenues to expand its powers; it is another thing altogether that its interventions risk the appearance of politicisation.

Evidence recorded from eSafety during the Baumgarten case revealed that the Commissioner sees herself as an “advocate for complainants”, to “assist them to achieve removal of material from social media platforms, and assisting in the enforcement of, or compliance with, providers’ terms of service”. This advocacy approach has created the appearance that censorship decisions are being driven by ideology.

Various legal disputes reveal the political nature of eSafety’s advocacy. In the Baumgarten matter, the reader will remember that eSafety initially succeeded in removing Ms Baumgarten’s post criticising the establishment of a queer club for young children. This post did not meet the threshold for adult cyber abuse, but eSafety sought and obtained its removal anyway, as was the case with the nearly 300 other adult cyber abuse cases informally finalised the same year. During the proceedings, Ms Baumgarten explained that she herself was bisexual and took issue with instructing young children about queer theory, regarding the informal action as “censorship of the types of gay people the eSafety Commissioner personally disagrees with”.

Following a similar pattern in 2025, the eSafety Commissioner issued a formal takedown order against Canadian activist Chris Elston, which was rejected on appeal at the Administrative Appeals Tribunal. Elston had criticised the suitability of the appointment of a transgender activist to a World Health Organisation panel of experts to advise on global transgender policy. Again, Mr. Elston’s post did not meet the legal threshold for takedown, suggesting the order may have been pursued for political reasons.

The scale of this shadow censorship regime is staggering.

This advocacy mindset extends beyond litigation to broader political interference, with the Commissioner openly positioning herself within national debates. In the lead-up to the referendum on the Voice, Ms Inman Grant said the vote was an “historic opportunity” and that the role of eSafety and government was to “make sure the online spaces are relatively safer spaces to be able to get this information out”.

That a bureaucrat would identify with a political cause like the Voice would be unremarkable. Indeed, the only jurisdiction to support the Voice was the ACT itself, the home of the Commonwealth public service. But what is alarming is that Ms Inman Grant’s office has the unique power to remove political content from the Internet, and the discretionary power to choose which complaints to investigate. The appearance of partisanship combined with discretionary power to censor is a dangerous combination.

Through legal overreach and partisan advocacy, the eSafety Commissioner has transformed what is already a problematic regulatory office into an instrument of political censorship. Her powers have been reined in only by the courts and only thanks to those with the resources to legally challenge eSafety’s edicts.

It is unlikely that the Albanese government will tolerate so many challenges to the Office of the eSafety Commissioner. In April 2026, Communications Minister Anika Wells announced the government would introduce legislation to update the OSA and introduce a “digital duty of care”, which would make eSafety substantially worse.

The “digital duty of care” was recommended in the OSA Review Report as a legal framework to place “responsibility on service providers to take reasonable steps to address and prevent foreseeable harms on their services”, with harms described elsewhere in the report as “harms to mental health” and “threats to social cohesion and national security”.

This could mean the introduction of a legal requirement for platforms to pre-emptively censor “harmful content” before it goes viral—a requirement enforced by the eSafety Commissioner. In this scenario, the censorship would be outsourced to social media companies, resulting in fewer appeal rights for Australian users. It could also mean greater scope for eSafety staff to impose their own values onto the regulatory framework.

The Office of the eSafety Commissioner represents an extraordinary but completely predictable failure of governance. It is what happens when sweeping powers are given to unaccountable regulators, with scant regard for the rule of law and broadly defined—and highly ideological—objectives. Until Parliament acts to rein in the Commissioner’s powers, Australians will remain subject to the political whims of a state sanctioned censor.

This article from the Winter 2026 edition of the IPA Review is written by IPA Research Fellow Margaret Chambers.