Summary

  • Australia's proposed following-only feed is not yet an enforceable entitlement. Its effectiveness depends on legislation, commencement of the proposed Determination and continuing compliance with requirements that make user preferences accessible, durable and meaningful.
  • The draft expressly permits advertising from accounts users do not follow. That permission establishes a commercial boundary around the proposed choice, without requiring advertisements or specifying their frequency, targeting or delivery.

A choice with two different meanings

Imagine a user opening a social-media service and deciding that the stream should contain material from accounts they have deliberately followed. The decision appears simple: replace a feed influenced by recommendations with one organised around an established set of relationships.

Yet there are two different promises concealed in that interface.

The first is a product promise. A platform can offer a setting, describe what it does and determine how reliably it operates. The user's experience depends on decisions made within the service: where the control appears, whether it is remembered and whether its practical effect matches its description.

The second is an institutional promise. A legal rule can define the choice, establish obligations surrounding its presentation and persistence, and provide a foundation for holding a service responsible when its conduct fails to meet the prescribed requirements.

Australia's “My Feed, My Way” proposal seeks to move part of this decision from the first category into the second.

On 15 October 2026, the Department of Infrastructure, Transport, Regional Development, Communications, Sport and the Arts opened consultation on the draft Online Safety (User Empowerment Tools) Determination 2026. Submissions are due by 23:59 Australian Eastern Daylight Time on 5 November 2026.

The published instrument carries a decisive warning: “DRAFT ONLY—NOT FOR SIGNATURE.”

That warning is more than administrative housekeeping. It separates what the government wants to require from what services are presently obliged to do under this particular proposal. The consultation offers a prospective architecture for user choice, not evidence that the architecture has entered legal operation.

The distinction establishes the central question for regulators, industry and users alike: under what conditions does an option displayed on a screen become a dependable entitlement?

The answer requires attention to three stages that public debate can easily compress into one. Parliament must establish the necessary legislative authority. The subordinate instrument must commence in accordance with its governing provisions. Services within its scope must then execute the requirements consistently enough for the promise to survive ordinary use.

There is also a fourth consideration, embedded inside the proposed option itself: what exactly does the entitlement protect?

The draft's answer is narrower than an escape from every mechanism through which platforms influence attention. It addresses a defined relationship between the content of a feed and the accounts a user follows. It does not establish that all selection, processing or commercial distribution disappears.

Most importantly, the proposed following-only feed can still include advertising from accounts outside the user's chosen network.

The measure therefore combines a substantial proposed change in user authority with an explicit limit on that authority. Understanding both elements is essential to judging its eventual significance.

The legal sequence comes before the interface

The institutional foundation is not a technical detail that can be postponed until implementation.

Section 2 of the draft Determination links its commencement to the commencement of Schedule 2 of the proposed Online Safety Amendment (Digital Duty of Care) Act 2026. Section 3 identifies subsection 27A(1) of the Online Safety Act 2021 as the relevant authority.

These references describe the legal architecture contemplated by the draft. They do not, by themselves, establish that Parliament has approved the relevant amendments or that the new determination-making power is operative.

On 15 October, Minister Anika Wells announced that the Online Safety Amendment (Digital Duty of Care) Bill 2026 had been introduced to Parliament. Introduction is a significant procedural step, but it is not passage, Royal Assent or commencement.

The distinction matters because a proposed subordinate instrument cannot acquire practical legal authority merely by citing provisions whose contemplated operation depends on legislation that has not been shown to have completed the necessary process.

The draft therefore expresses a conditional regulatory design.

First, the legislative basis must be validly established. The provisions necessary to support the proposed determination must enter into force in the manner ultimately enacted.

Second, the determination itself must be validly made and commence under the applicable legal arrangements. Section 2's reference to Schedule 2 is important precisely because it makes the contemplated timing dependent on another legal event.

Third, the requirements must actually be implemented by the services to which they apply.

This sequence prevents three different claims from being mistaken for one another: that the government has proposed a requirement, that an enforceable requirement exists and that affected users can reliably exercise it.

The government has advanced the first proposition. The facts available on 16 October 2026 do not establish the second or third.

The proposed entitlement must consequently be assessed in the conditional tense. It would change the legal position if the necessary legislative and subordinate steps occurred and the relevant obligations took effect. Until then, the draft identifies the government's intended direction rather than a completed alteration of platform duties.

Such precision is not a pedantic obstacle to understanding reform. It is fundamental to institutional legitimacy. Public authority is credible when the obligation claimed matches the authority legally conferred.

What the following-only feed would actually require

The draft does not attempt to regulate every digital service or every mechanism for organising online content.

Section 6 concerns specified social-media services and platforms that provide user content through recommender functions. Its provisions distinguish algorithmically profiled content from a proposed following-only alternative.

The principal idea is that users within the relevant scope should be able to choose a feed in which ordinary content comes from accounts they follow, rather than having unrelated accounts introduced through the recommendation system.

This is a change in the permitted source of material, not necessarily a change in every process applied to that material.

A following-only feed should not automatically be described as chronological. The provisions supplied do not establish an obligation to order posts solely by publication time, nor do they prohibit all ranking or processing of eligible content.

This is an important distinction.

A rule controlling which accounts may supply ordinary feed material addresses one source of platform influence: the introduction of posts from outside the user's declared social relationships.

A rule requiring chronological display would address another: the ordering of material within the allowed set.

The proposed Australian measure should not be credited with the second merely because the first might make the resulting feed feel less recommendation-driven.

Nor does following an account necessarily mean that every item from that account must appear. The entitlement described in the draft concerns a defined feed option, not a guarantee of exhaustive or uninterrupted delivery of every possible post.

The relevant legal boundary is therefore the distinction between ordinary material from followed accounts and ordinary material from accounts outside that relationship.

This matters because a user can reasonably understand “following-only” as a promise about the provenance of ordinary content, while still expecting the service to undertake some processing necessary to present the feed.

A regulator evaluating such a system would need to preserve that distinction. The question would not simply be whether the platform uses an algorithm. It would be whether the operation of the chosen feed respects the obligations actually imposed.

That makes the proposal more precise, and potentially more verifiable, than an undifferentiated campaign against algorithms.

It also makes public description unusually important. If the option is promoted as eliminating all platform mediation, expectations could exceed the legal protection contemplated by the text.

The stronger interpretation is also the narrower one: the government proposes to give users a defined choice about the ordinary content sources included in a feed, while leaving other forms of processing outside that particular promise unless separately restricted.

Why the durability of the setting matters

Section 5 is the operational centre of the proposal.

An entitlement to choose is of limited value if the relevant control is hidden, difficult to use or quietly displaced as a user moves through the service.

The draft addresses this problem by proposing several linked requirements. User-empowerment tools should be obvious, easy to use, prominent, available and functional. They should not make service use or navigation more difficult, directly or indirectly discourage their own use, or produce an experience that is materially inferior to the alternative.

The instrument also proposes that settings persist during and between visits unless the user expressly changes them.

These conditions turn what might otherwise be regarded as minor interface details into potential elements of compliance.

Prominence concerns the ability to discover the choice. Ease of use concerns the burden of exercising it. Comparable experience concerns whether exercising the choice imposes a disadvantage unrelated to the user's stated preference. Persistence concerns whether the service continues to respect the decision after the immediate selection has been made.

The provisions form a coherent institutional argument: a choice must not merely exist; it must remain usable without repeated negotiation.

Consider the difference between two hypothetical designs.

In one, a user selects a following-only option, can readily understand its operation and encounters the same preference after returning to the service.

In another, the user must repeatedly locate the option, navigate additional steps or restore the preference after routine visits.

Both designs might display a button labelled “Following.” Their implications for user control are very different.

These examples are analytical illustrations, not findings about the conduct of any identified service.

The proposed requirements aim to prevent the formal availability of a control from substituting for meaningful access to it.

Persistence is especially consequential because it reallocates the burden of repetition. Without a durability requirement, a platform can retain substantial practical influence simply because the user must keep expressing the same decision. With a binding persistence rule, the selected preference becomes the reference point until the user expressly changes it.

A continuing choice is institutionally different from a temporary exception.

The comparability requirement raises an additional question: what does it mean to provide a comparable experience when the user has deliberately selected a different information environment?

Comparability cannot logically require identical content. The option exists because the content sources are meant to differ.

Instead, the requirement points toward the conditions of use: users should not have to accept avoidable functional disadvantages simply because they exercise a protected preference.

The eventual compliance question would depend on the precise legal wording and the circumstances of implementation. The draft does not, on the supplied evidence, establish a completed test or a record of actual compliance.

Nevertheless, its design recognises an important feature of digital power. Control over an interface can determine whether an apparently voluntary choice is realistic.

Legislation can prescribe the existence of a right; interface design determines how frequently users can exercise it without friction.

The advertising exception defines the commercial frontier

Section 6(6) introduces the most revealing limitation.

The draft states that a following-only feed may include advertising material, including advertising from accounts the user does not follow.

The word “may” is decisive.

It permits this category of advertising to appear. It does not require a service to insert advertisements, prescribe a particular advertising load or compel exposure to promotions from unfamiliar accounts.

Neither does the paragraph, on the facts supplied, establish rules for advertising frequency, total volume, targeting criteria or auction design.

It would therefore be incorrect to infer that the proposal mandates a certain quantity of advertising, changes the way advertising auctions operate or establishes a new targeting system.

It would also be incorrect to infer that advertising is generally unregulated merely because this specific provision allows it within the following-only feed.

The provision addresses the scope of this proposed choice. It is not a comprehensive statement of every law that may govern advertisements.

But the permission is still institutionally significant.

The following-only option proposes to restrict one route through which accounts outside a user's chosen network enter the stream: ordinary material delivered through the feed.

At the same time, it leaves open another route: advertising material.

The distinction means that a user's decision not to receive ordinary posts from unfamiliar accounts would not necessarily prevent paid material from those accounts appearing alongside followed content.

This is more than a drafting footnote. It identifies the perimeter of the proposed entitlement.

A user choosing a following-only feed would acquire a particular form of control over ordinary content sources, rather than an absolute right to exclude all accounts outside their network from the screen.

That difference has commercial significance because the platform's advertising arrangements occupy a distinct position within the proposed design.

The draft does not demonstrate that any platform has acted in a particular way in response to the exception. Nor does it establish that advertising exposure will increase, decrease or remain constant.

What it does establish is an asymmetry of permission.

Ordinary material from an account outside the user's followed network is constrained by the following-only design. Advertising from such an account remains permissible.

The line is drawn by the status of the material, not simply by the relationship between the user and its source.

This gives the choice an unusual double character. It would narrow the system's authority to introduce certain ordinary posts while preserving a channel through which the commercial system may introduce material from outside the same network.

The boundary deserves greater attention than a broad claim that users will be able to “switch off the algorithm.”

The proposed choice cannot accurately be described as a complete withdrawal from platform-curated exposure. Nor should the advertising exception be exaggerated into proof that the choice has no value.

Both propositions obscure the legal structure.

Control over ordinary recommendations may matter considerably to users who prefer familiar content sources. An advertising permission limits that control without necessarily making the remaining protection meaningless.

The serious question is whether the distinction between ordinary content and advertising can be made intelligible, verifiable and consistent with the wider promise of user empowerment.

That is where the commercial boundary becomes a question of institutional design.

The default matters, but only when the rule begins

Section 7 addresses users who have not already expressed a relevant feed preference.

Under the proposal, where an authenticated user has not selected a preference at the time the rule commences, the service must offer the choice when that user next accesses the service.

This provision recognises that an option may remain practically invisible if users are left to discover it unaided.

It would create a prospective duty to present a choice at a defined point for a defined category of users.

Its conditions are important.

The provision concerns an authenticated user. It addresses the absence of an existing preference when the rule commences. It specifies the next access as the occasion on which the choice is to be offered.

It does not establish that every user must immediately be moved to a following-only feed. Nor does it establish that the obligation applies on 16 October 2026.

The proposed rule concerns presentation of an available decision, not an automatic determination of what every user should prefer.

This distinction protects the underlying legitimacy of the measure. An instrument intended to empower users should avoid replacing one unchosen information environment with another merely because policymakers regard the second as preferable.

The regulatory objective is to establish a meaningful opportunity for choice, followed by respect for the preference expressed.

In combination with Section 5's persistence requirement, Section 7 would help create a sequence in which the service presents the decision, the user exercises it and the service preserves the result.

Such a sequence is more substantial than simply requiring a settings menu.

It begins to describe an ongoing relationship between a user's declared preference and the platform's obligations.

From political responsibility to institutional authority

Minister Wells has framed the broader Digital Duty of Care agenda in terms of platform responsibility, including in her National Press Club address.

That framing helps explain why the government is pursuing user-empowerment tools as part of a wider legislative programme. It is a ministerial account of policy purpose, however, not independent evidence that any particular obligation has already taken effect.

The distinction between political justification and binding authority is especially important here because the proposed determination depends on the legislative process.

Parliament's role is not interchangeable with the minister's announcement. The eventual legal instrument is not interchangeable with an exposure draft. A platform's actual implementation is not interchangeable with either.

Each stage has a different institutional function.

Political leadership proposes a direction. Parliament determines the legislative foundation. The subordinate rule specifies requirements within the authority available to it. Services must then undertake the operational work necessary to make those requirements effective.

Public legitimacy depends partly on whether those responsibilities remain visible.

If a government announces a choice before the relevant obligations exist, users may mistake a proposal for an entitlement. If a platform later makes a formally compliant option difficult to exercise, the legal existence of the option may offer less practical protection than the public language suggests.

Conversely, clear requirements concerning accessibility and persistence could make the implementation question easier to examine than vague promises about responsible algorithms.

The proposed Australian approach is therefore best understood as a move toward the governance of user choice itself.

Rather than determining every item that should appear in a feed, the draft seeks to define a decision that users should be able to make and the conditions under which services must respect it.

It also preserves a commercial exception within the same decision.

That combination is the proposal's central institutional compromise.

The state would seek to establish a protected channel of user direction over ordinary feed sources without making the following-only option synonymous with exclusion from advertising.

Whether the result becomes enforceable, effective and trusted depends on events that had not been established as complete on 16 October 2026.

For now, the proposal should be judged as a carefully bounded attempt to convert a platform setting into a continuing obligation—not as a completed transfer of control over the entire feed.

Sources

  1. Department of Infrastructure, Transport, Regional Development, Communications, Sport and the Arts — Choice of feeds on social media: public consultation. Consultation opened 15 October 2026; submissions close 5 November 2026 at 23:59 AEDT.

  2. Australian Government — Draft Online Safety (User Empowerment Tools) Determination 2026 (PDF). Primary instrument for proposed commencement, authority, usability requirements, following-only feeds, advertising permission and initial choice.

  3. Minister Anika Wells — Digital Duty of Care: Next step in Australia's world-leading reforms. Official announcement dated 15 October 2026 concerning introduction of the Digital Duty of Care Bill.

  4. Department of Infrastructure — Exposure Draft: Online Safety Amendment (Digital Duty of Care) Bill 2026. Earlier exposure-draft legislative material providing context for the proposed statutory framework.

  5. Minister Anika Wells — Address to the National Press Club. Ministerial explanation of the government's platform-responsibility and Digital Duty of Care policy framing.