Supplementary submission on the Online Safety Amendment (Digital Duty of Care) Bill 2026

Executive summary

My recommendation remains that the Bill should not pass. This supplement addresses the amendments adopted by the House of Representatives on 28 October 2026, after my original submission.

The amendments broaden retrospective validation from shortened consultation periods to failures to publish draft codes, invite submissions or comply with the consultation requirements for standards. The Government explains that it wants continuity during the transition to the new duty. That is a reason to examine the problem, not a sufficient explanation of which instruments and legal consequences Parliament is being asked to validate. The committee should obtain that information before recommending any validating provision.¹²³

The new emotional-intimacy and dependence category also needs scrutiny. The new requirements for usable user-empowerment tools are an improvement and should be retained in targeted legislation. Neither change resolves the objections to the general duty. The amended Bill and its explanatory material also contain specific errors which should be corrected.

Contents

• 1. Scope
• 2. Retrospective validation
• 3. Emotional intimacy and dependence
• 4. User-empowerment tools
• 5. Drafting and explanatory consistency
• 6. Recommendations

1. Scope

I make this supplement in my personal capacity. It supplements my submission of 26 October 2026 and does not replace it. It compares the introduced Bill with the text as read a third time on 28 October and considers the Government’s supplementary explanatory memorandum.⁴¹²³

I also correct one detail in section 12 of my original submission. I described the fake-nude removal measure as applying to services predominantly designed or used for that purpose. Introduced section 86B(1) required both conditions. The House amendments changed that conjunction to an alternative so either condition now suffices. My recommendation for accurate targeting, reasons and review remains unchanged.¹²³

This supplement concentrates on changes material to my original recommendations. It does not repeat the original discussion of service scope, cost, age assurance, encryption, research powers or consultation arrangements.

2. Retrospective validation

The introduced items 90 and 91 protected validity where a consultation period was shorter than required. The amended item 90 would also protect an industry code despite a failure to publish its draft or invite submissions. Amended item 91 covers failure to comply with section 148, which governs public consultation on industry standards. Both retain the retrospective effect that validity is taken never to have been affected.¹²

That is a substantive expansion. A shortened opportunity to comment and no opportunity to inspect or respond to a draft are different procedural failures. Parliament should not treat them as interchangeable technical defects.

The supplementary memorandum states the Government’s reason: preserving the validity of codes and standards during the transition to the duty and limiting gaps in protection. I recognise that continuity matters. Removing an effective protection without a replacement can harm the people who rely on it. The memorandum nevertheless does not identify the affected instruments, the particular failures or their consequences.³

Before recommending validation, the committee should require the Government to identify each instrument said to need it, the requirement not met, what occurred and the effect on public participation, enforcement, liability and any known proceedings. It should also ask why the problem cannot be corrected through lawful consultation and remaking or another narrowly drawn continuity measure. If the Government cannot identify an actual defect, it should explain why this broad retrospective provision is necessary.

I do not assume that any particular instrument is invalid. The issue is that Parliament is being asked to remove specified grounds on which validity could be challenged without an instrument-specific account of the problem. A general assurance that the provisions maintain protection does not answer what legal consequences they would change.

Schedule 1 commences the day after Royal Assent. The duty and repeal provisions in Schedules 2 and 3 share the later commencement, after the twelve-month period. Validation would therefore take effect well before the replacement regime. A transition rationale should not obscure its immediate retrospective operation.²

My recommendation remains that Part 5 be removed unless the Government supplies that account and justifies a tightly confined measure. Any necessary continuity arrangement should preserve identified protections without validating unrelated failures or consequences which Parliament has not examined.

3. Emotional intimacy and dependence

New paragraph 25E(1)(da) covers material or conduct which simulates emotional intimacy and encourages or fosters a child’s dependence on it in place of human relationships. It is not confined to artificial-intelligence companions. Its wording requires the intimacy and dependence elements together, which should be recognised when assessing its reach.²

A service can exploit a child’s vulnerabilities, pressure them to remain engaged and discourage support from other people. Providers should answer for such conduct within their control. That does not make every simulation of intimacy or form of reliance a sufficiently precise category for a general statutory duty.

The paragraph does not define emotional intimacy, dependence or replacement of human relationships. It contains no separate likelihood or seriousness threshold. The general proportionality considerations in section 26A and the excluded-material provisions remain relevant but they do not give the new category a clear boundary. Providers could respond conservatively by restricting beneficial support or collecting more information about a child’s relationships to establish what dependence means in that case.

The committee should require evidence of the conduct being targeted and explain how the category distinguishes exploitative design from legitimate assistance, creative use and confidential health or support information. Children with disability and children in unsafe households need particular consideration. A protection should not depend on disclosing intimate circumstances to a platform or obtaining approval from an unsafe parent.

Schedules 2 and 3 should still be withdrawn. If Parliament later pursues a targeted measure, it should identify provider-controlled conduct creating a reasonably foreseeable risk of serious harm and protect confidential support, privacy and access. A foreseeable risk can justify preventive action before injury occurs. The restriction still needs a defensible connection to that risk.

4. User-empowerment tools

New section 27BA requires tools to be accessible, easy to use, prominent, functional and available on an ongoing basis. Their use must not affect general service functionality or reduce or impair user experience. The requirements have a specific civil penalty and are included in the remedial-direction framework. These are useful additions.²

A control hidden behind repeated prompts, difficult to operate with assistive technology or followed by loss of unrelated functionality is not meaningful user choice. I support retaining these protections in targeted legislation. My separate submission on the User Empowerment Tools Determination addresses the detailed operation of feed controls.⁵

Section 27BA does not itself specify which tools a service must offer; that remains dependent on section 27A and the relevant instrument. Availability on an ongoing basis is also different from preserving the user’s selected setting. The instrument should require persistence across ordinary sessions, updates and access methods, straightforward reversal and testing with keyboard and assistive-technology users. Compliance should be demonstrable without additional profiling.

The no-impairment wording should prevent retaliation and coercive friction. It should not treat the user’s intended change as an impairment. A person choosing a following-based feed expects different recommendations. A person disabling an engagement feature expects it to stop operating. The legal test should protect that choice rather than require the alternative to reproduce the experience the user rejected.

The prohibition in section 27A(4) on considering particular material when making the instrument should remain. User control over a service does not require a government preference about which lawful speakers or viewpoints the user should encounter.²

5. Drafting and explanatory consistency

Section 25F(2) still refers to subparagraph (1)(c)(iv), although the substituted paragraph (1)(c) contains only (i) and (ii). The film and computer-game provision is now in (ii). The cross-reference should be corrected if the intention is to retain the exception for user-generated game material. This is a specific drafting error, not a claim that the whole Bill is invalid.²

The supplementary memorandum also needs correction. On page 9 it describes paragraph 25F(1)(c)(ii), an excluded-material provision, as extending material into seriously harmful material or conduct. The operative text does not do that. Section 25F defines exclusions and section 25D(2) expressly excludes class 2 material from seriously harmful material or conduct. Whether material is harmful to children remains a separate question under section 25E. The explanation should state those distinctions accurately.²³

These corrections would improve the text and its explanation. They would not resolve the substantive concerns about the duty or justify the expanded validation provision.

6. Recommendations

I recommend that the committee maintain scrutiny of the general duty and:

1. Recommend that the Bill not be passed, with Schedules 2 and 3 withdrawn and targeted measures considered separately.
2. Remove Schedule 1 Part 5 unless the Government identifies the instruments, consultation failures and legal consequences requiring validation and justifies a narrowly confined response.
3. Require separate justification of the new dependence category against evidence, provider control, serious foreseeable harm, privacy and beneficial access.
4. Retain the useful section 27BA protections in targeted legislation, with persistent user choice, accessible operation and a clear distinction between intended changes and retaliatory impairment.
5. Correct section 25F(2) and the supplementary memorandum’s explanation of excluded material before any passage.

References

1. Parliament of Australia, Online Safety Amendment (Digital Duty of Care) Bill 2026: first-reading text, 15 October 2026, Schedule 1, items 5, 90 and 91, printed pp. 5 and 28. ↩ ↩² ↩³ ↩⁴

2. Parliament of Australia, Online Safety Amendment (Digital Duty of Care) Bill 2026: as read a third time, 28 October 2026, commencement table, Schedule 1, items 5, 90 and 91 and Schedule 2, proposed ss 25E, 25F, 27A, 27BA and 27C, printed pp. 2, 5, 28, 39 to 41 and 48 to 49. ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰

3. Parliament of Australia, Supplementary Explanatory Memorandum: amendments and new clauses to be moved on behalf of the Government, 28 October 2026, especially pp. 3, 6 and 8 to 9. This memorandum concerns the Government amendments rather than every amendment adopted by the House. ↩ ↩² ↩³ ↩⁴ ↩⁵

4. Zen Dodd, Submission on the Online Safety Amendment (Digital Duty of Care) Bill 2026, 26 October 2026, especially sections 6, 9, 12 and 14. ↩

5. Zen Dodd, Submission on the Online Safety (User Empowerment Tools) Determination 2026, 1 November 2026. ↩