Corporate boardroom or negotiation table representing big tech pressure on Australian copyright law
| | |

Big Tech Told Australia to Rewrite Its Copyright Laws. Or Else.

This week, the four most powerful AI companies on earth flew representatives to Canberra to tell a Senate committee that Australia needs to change its copyright law. Or AI investment will go elsewhere.

Google said an opt-in copyright regime would make AI development “impossible.” Microsoft said legal uncertainty could drive investment away. OpenAI and Anthropic both turned up to make the same case in different registers. The message was consistent: rewrite the rules in our favour, or watch the money leave.

Senator David Pocock called it what it was. Big tech, he said, was “dictating terms to every single Australian.”

He wasn’t wrong.


Here’s what was actually being asked. Right now, if an AI company wants to train a model on a writer’s work, a musician’s recording, or an Australian news article, they need permission. Without it, they’re in breach of copyright. The companies want that to change. They want a broad exception that lets them scrape and train on any content they can access, unless the creator explicitly opts out. Anthropic’s Jeff Bleich put it bluntly: “you can’t license the entire internet.” So copyright must change.

The argument is that licensing is too hard, so the law should remove the requirement to license.

Think about that for a second. These are trillion-dollar companies. If anyone has the resources to build licensing infrastructure, it’s them. The Writers’ Guild made exactly this point at the inquiry. Creators want control over their work, not just payment after the fact. “Give us opt-out rights” is a very different ask from “let us build a system where you can be compensated.”


The music industry was less diplomatic. ARIA, the peak body for Australian music, said artists will be “roadkill” if Australia hands big tech what it wants. They went further: Australia is being used as a “Trojan horse,” a test case to build the international precedent that AI companies can train on creative work without consent.

That’s a pointed thing to say. But the timeline supports it. Multiple countries are watching how Australia handles this. If Australia legislates a broad copyright exception, it becomes the model others are pressured to follow.


The ABC, which creates a substantial amount of publicly funded content, took the opposite position from Google. Existing copyright law doesn’t need changing, the ABC told the inquiry. The current framework is capable of handling AI. The problem is enforcement, not the statute.

Then came the moment that should have received more coverage than it did.

Trellis Data, an Australian AI lab backed by Main Sequence Ventures, told the inquiry that competitive AI models don’t actually need a copyright carve-out. That contradicts the core of big tech’s argument: the claim that you simply cannot build capable AI without training on copyrighted content without restriction. Trellis says you can. Their team is building it.

One Australian AI company, not a trillion-dollar multinational, just punched a hole in the “impossible” framing Google opened with.


To be fair about what’s actually happening here: AI companies genuinely do face a difficult copyright field. Training data provenance is complicated. Blanket licensing across millions of works is hard. These are real problems, not invented ones.

But the solution being proposed is not a neutral technical fix. “Change the law so we don’t have to deal with this” is It’s a transfer of economic rights from creators to platforms. The writers, musicians, and journalists whose work would be used as training data lose the right to say no, and they lose use over what compensation looks like.

The ABC says the existing law can handle this. Trellis Data says you can build competitive AI within copyright constraints. Together they suggest the “impossible” framing is negotiating strategy, not engineering reality.


For Australian small businesses, this matters in a way that isn’t immediately obvious.

The content your business creates: blog posts, product descriptions, how-to videos, client case studies. It could be swept into AI training data under any broad exception. You wouldn’t be asked. You wouldn’t be compensated. If you wanted to opt out, you’d need to figure out how, platform by platform, and hope the mechanism actually works.

This is also about the competitive position of Australian creative industries, which are real customers of small businesses. Graphic designers, photographers, copywriters, musicians. If their ability to control their work and charge for it is eroded, that hits the businesses that hire them.

There’s a longer read on what this copyright fight means for Australian small businesses here.


The inquiry continues. Final recommendations aren’t due until November 30. The big tech companies have lobbyists in Canberra and relationships with government that independent creators do not. That asymmetry is worth keeping in mind when the recommendations land.

Senator Pocock’s framing was accurate. What happened this week was not a technical policy debate. It was a negotiation, and one side had a lot more use than the other.


📬 The SmallBizAI Brief

One practical AI tip for Australian small business. Every Tuesday. Free.

Join business owners getting smarter about AI every week.

Subscribe free →

Sources

More hot takes: Hot Takes →

📖 More deep dives: News Deep Dives →

Similar Posts

Leave a Reply

Your email address will not be published. Required fields are marked *