FACT-CHECK: Would the Digital Duty of Care Bill force religious teaching offline?
As Labor’s online safety overhaul moves through the Senate, a Sydney Anglican bishop has warned it could silence the churches. Bishop Michael Stead — who chairs the Christian legal body Freedom for Faith and the Sydney Anglican diocese’s Religious Freedom Reference Group — told a Senate committee on 10 November that the Online Safety Amendment (Digital Duty of Care) Bill 2026 could see orthodox Christian teaching on marriage and gender treated as harmful content and taken offline. The House passed the bill on 28 October, and the government wants it through the Senate by the end of November. We checked his claims against the bill.
What the bill actually says
The bill replaces Australia’s online safety codes with a single statutory duty. Under section 26(1), a person responsible for an online service “must ensure, so far as is reasonably practicable, a safe online environment.” It’s a systemic duty: it asks whether a platform’s design and systems manage foreseeable risk, and it runs against those systems rather than against any individual post. A breach carries a civil penalty of up to about $109 million for a company.
The duty applies to defined categories. Section 25D sets out an exhaustive list of “seriously harmful material and conduct” (and adults and children alike must be protected from it) covering child sexual abuse material, grooming, promotion of suicide or self-harm, incitement of extreme or sexual violence, explicit threats and serious harassment. One category matters more than the rest for this debate: section 25D(1)(h) covers “material or conduct that encourages, promotes, incites or instructs gender-based violence”, and the bill defines gender-based violence in section 5 as “any form of physical or non-physical violence, harassment, abuse or threats, based on gender”. The bill does not define “gender”.
After the exposure draft drew concern about over-moderation, the bill was amended to carve several categories of content out of the harm definitions: professional news, material produced for a literary, artistic or educational purpose that is not contrary to the public interest, scientific and academic research, and material in the public interest. There is no religious exclusion.
Claim 1 — “Church teaching could be classified as gender-based violence and taken offline.”
Verdict: Overstated, but with a kernel of truth
The strong version doesn’t hold water. The bill names no religious teaching and creates no offence aimed at religion. Whether teaching that men are the head of the household is “gender-based violence” is a matter for debate, not something the bill sets out to capture, and the duty runs against a platform’s systems in general rather than giving anyone the power to order a particular sermon down. The eSafety Commissioner, Julie Inman Grant, told senators the fear was “a stretch”, and that eSafety’s focus was millions of high-harm targets, not churches.
But the kernel is stronger than a tidy dismissal allows. Gender-based violence is an express category of seriously harmful material; its statutory definition reaches non-physical “abuse” and “harassment” based on gender; and “gender” is left undefined (though technically not entirely unbounded: the Acts Interpretation Act 1901 (s 23) provides that in any Commonwealth Act, words importing a gender include every other gender).
Stead’s submission makes the specific point that teaching on male headship in marriage and the church is “characterised by some critics as promoting ‘control’ or as a form of abuse”. Nothing in the bill says that reading is right, but nothing rules it out either, and it'll be a platform, not a court, that makes that first judgement.
Add a nine-figure penalty and the standing incentive is to over-remove rather than risk it, a concern the United States government itself raised in its submission on the draft. So the risk Stead points to is real: it runs partly through a vague and undefined category in the bill, and partly through the fines that make platforms cautious. But it's not a bill that sets out to define religious teaching as harmful.
Claim 2 — “‘Harmful’ is undefined, leaving the line to the government of the day.”
Verdict: Overstated.
Two things are wrong with it. First, “undefined” overstates the position: the bill sets out closed, exhaustive lists — seriously harmful material in section 25D, material harmful to children elsewhere — tied to a section 5 definition of “serious harm”. The lists have fuzzy edges, “serious distress” and the undefined “gender” among them, and that vagueness is a fair criticism. But vague is not the same as undefined, and the categories are in the bill for anyone to read.
Second, the “government of the day” point has been overtaken by the bill’s own history. The exposure draft did let the minister add new categories of harm by legislative instrument: the very mechanism that would have left the line to the government of the day. That power was removed before the bill was introduced.
Announcing the change on 15 October, Communications Minister Anika Wells said the government had “removed the ministerial rule-making power to define new harms”, and that new harms could be added only by Parliament amending the Act. Both harm lists are now closed and can be expanded only by primary legislation.
What survives of Stead’s concern is narrower: a government agency, eSafety, writes the guidance platforms follow, so the executive still shapes how the duty works in practice — but it cannot quietly expand what counts as harm, which is what his claim asserts it can.
Claim 3 — “The bill gives religious expression no protection.”
Verdict: Accurate.
The bill carves professional news, educational, artistic, research and public-interest material out of the harm definitions under section 25F. It contains no equivalent carve-out for religious expression, and none appeared in the exposure draft either. Freedom for Faith has asked for one to be added.
This is the fact that keeps Claim 1’s kernel alive: if platforms do over-remove to stay safe, religious content has no express statutory shield to point to, as a news publisher’s news material does.
The bottom line
On the bill as written, the alarming version of the claim — that it targets or defines religious teaching as harmful — doesn’t stand up: it names no such teaching, and the serious harms it lists are criminal and safety categories. But the narrower case is stronger than the government’s reassurances suggest. One of those categories, gender-based violence, is defined broadly and loosely enough — non-physical, resting on an undefined “gender” — that whether it reaches contested religious teaching is a question platforms will answer first, under threat of considerable fines and with no religious carve-out to fall back on.
Stead’s weakest point is the claim that a future government could expand the harms at will: the power that would have allowed that was taken out before the bill reached Parliament.
Whether the vague category and the missing carve-out warrant a religious exclusion, as Freedom for Faith argues, or are adequately handled by the bill’s limits and the Commissioner’s assurances, as the government says, is the question now before the Senate.
This briefing reflects the position as at date of publication and is general information, not legal advice. It rests on the primary sources linked above; where a provision’s meaning is genuinely contested, we say so.