AI copyright Australia: Pocock and Ayres clash in Senate

David Pocock has forced Australia’s AI copyright fight into open politics, accusing the Albanese government of weighing changes that could let AI companies train models on local content on looser terms. In a Senate exchange on 23 June, Industry and Science Minister Tim Ayres rejected the claim and said copyright protections would not be weakened.

Canberra’s argument now reaches beyond question time. Publishers, artists and AI vendors are watching for any sign that the government is revisiting training-data rules, because even a narrow change to copyright law could decide who must pay for access to local work. Pocock used parliamentary privilege to put the allegation on the record. Ayres tried to pull the debate back to what ministers have already said in public.

Ayres’ clearest answer was brief: “There will be no undermining of copyright protections,” according to the Senate transcript.

Pocock said that did not answer his question. In comments to ABC News, the independent senator said Ayres had accused him of reckless speculation while still refusing to rule out a live policy discussion. His question centred on a claimed 15 July 2026 timetable for a possible announcement, a detail not confirmed elsewhere on the public record.

“The minister accused me of reckless speculation, but, rather curiously, could not simply rule out the speculation,” Pocock told ABC News.

That timetable explains why Ayres pointed back to the government’s earlier position. In the Senate, he cited an attorney-general statement from 26 October 2025 that ruled out a text-and-data-mining exception under Australian copyright law. The Guardian reported that Ayres had also said on 6 August 2025 that the government had no plans to make changes. If ministers are examining the issue again, they are doing so against a public record that says no such shift is planned.

For rights holders, the argument is practical. A carve-out or looser exception would influence whether publishers and other content owners can insist on licences before their work is used to train models, or whether developers get a broader statutory path to use that material without prior deals. The policy difference is not academic for newsrooms, software firms or artists whose work can be scraped at scale. That is why a private consultation dispute has become a visible political fight.

Assistant minister Andrew Charlton added to the sensitivity in a recent ministerial press conference, saying “Australian data is the juice in the machine of AI.” The remark did not announce a copyright carve-out. It did show how closely data access now sits to Australia’s wider AI ambitions as ministers court investment in local computing and data-centre capacity.

No draft law has been published and no copyright change has been confirmed on the record. For now, the dispute is about disclosure as much as policy. Once the issue is aired in the Senate, it becomes harder for ministers to treat AI training data as a closed-door copyright question rather than a live technology policy test.