Australia Weighs Copyright Changes for AI: What Legal Teams Should Prepare For
Date: October 26, 2025 (GMT+8)
Australia will host a two-day meeting on Monday to consider targeted updates to its copyright framework in light of AI. The Copyright and AI Reference Group will workshop options that balance innovation with fair payment for creators.
The government flagged three focus areas. First, a potential paid collective licensing framework for AI users under the Copyright Act. Second, lowering the cost of enforcement actions. Third, clarifying how copyright applies to AI-generated outputs.
Text and data mining will not be carved out: the government has ruled out a TDM exception that would allow free training on local works. "There are no plans to weaken copyright protections when it comes to AI," Michelle Rowland said in a statement. "The tech industry and the creative sector must now come together and find sensible and workable solutions to support innovation while ensuring creators are compensated."
Collective Licensing for AI Training: What It Could Mean
A paid collective license could centralize permissions for training datasets and potentially standardize pricing and terms. Expect debate over scope (training vs. fine-tuning), audit rights, data provenance requirements, and how exemptions (e.g., truly public domain) are treated.
For in-house counsel, the likely issues to plan for are budget impact, proof-of-license workflows, and how any scheme interacts with existing licenses your company holds. Vendors that provide models or datasets may need to certify licensed sources, with warranties and indemnities to match.
No TDM Exception: Practical Implications
Without a statutory TDM exception, unlicensed ingestion of copyright-protected works for training remains risky. Fair dealing in Australia is narrow and purpose-based; it does not generally cover commercial AI training.
Legal teams should pressure-test data sourcing: maintain a catalog of datasets, their licenses, and any opt-out compliance. If relying on vendors, secure representations about lawful sourcing and keep audit rights alive.
Lower-Cost Enforcement: Risk and Opportunity
If enforcement becomes cheaper, expect more actions from individual creators and smaller publishers. This could increase exposure for businesses using unvetted models or content pipelines.
On the other hand, rights holders may find it easier to act against clear misuse. Clear takedown and settlement playbooks will help on both sides.
AI-Generated Outputs: Ownership and Protection
The group will consider how copyright applies to AI-produced material. Until further guidance lands, treat fully automated outputs as legally uncertain for protection and exclusivity.
Where possible, document human input that is specific, creative, and determinative of the final output. Contract terms with clients and vendors should reflect that some outputs may have thin or no copyright protection and may require confidentiality or contractual control instead.
Action Checklist for General Counsel and IP Teams
- Inventory all AI use cases (training, fine-tuning, inference) and map rights exposure for each.
- Establish a licensing strategy for datasets and model providers; centralize approvals and record-keeping.
- Update procurement: warranties on lawful data sourcing, indemnities, audit rights, and incident notification.
- Set content provenance standards: source tagging, logs for training/fine-tuning, and model/version traceability.
- Review marketing claims about originality and exclusivity of AI outputs; add disclaimers where needed.
- Prepare an enforcement playbook: intake, evidence capture, quick evaluation, and settlement ranges.
- Budget for potential collective licensing fees; model scenarios by product line and data volume.
- Train product, data, and engineering teams on permissible sources and license terms.
- Monitor government releases and be ready to submit feedback during consultation stages.
Key References
Government materials and the Copyright Act will be the primary sources as proposals emerge. For updates, see the Attorney-General's Department copyright page and the current Copyright Act:
Upskilling Your Legal Team
If your organization is standing up AI governance or contract standards, structured training can accelerate adoption and reduce missteps. See role-based options here: AI courses by job.
Your membership also unlocks:
