OpenAI and Anthropic put their copyright asks to Australia in writing

Share
OpenAI and Anthropic put their copyright asks to Australia in writing

Australia's copyright fight stopped being a rumour this morning: the two labs asking Canberra to rewrite its law have now filed their arguments where anyone can read them.

OpenAI and Anthropic have filed submissions to parliament's Joint Select Committee on Artificial Intelligence setting out what they want Australia's copyright regime to be — and both tie it to whether they train and build in the country. OpenAI asked for a "balanced copyright framework" that lets models learn from publicly available information, leaves room for deals with rights holders, and keeps protections against pirated material and infringing outputs; it did not say whether creators should be able to opt out, or whether developers should have to pay. Anthropic conceded more ground: it told the committee it has accepted it will not get US-style training exemptions, but that the government could instead create a payment system — one it says would unlock large infrastructure investment and pull Australian AI researchers home from Silicon Valley. The submissions were made public on Tuesday. Both companies argue that legal certainty determines where models get trained and where the computing that serves them gets built, which is the leverage they are actually spending.

The backdrop is a consultation document the government would clearly rather not be discussing in public. Screenshots of proposals titled "AI on Australian Terms," tabled in the Senate by independent senator David Pocock last week, set out two options for letting AI firms train on Australian material by default — one conditional on them striking deals with rights-holder organisations and collecting societies, the other gated on digital opt-outs. Under both, creators who want out would have to "digitally protect" their work. Shadow attorney-general Michaelia Cash called it "a text and data mining exception with an opt-out bolted on," which is precisely the design the government said it had ruled out. The music licensing body APRA AMCOS put the objection shorter: "investment that is conditional on a country rewriting its property law is not investment." We covered the parallel standards consultation last week — Australia's AI standards: report rogue incidents, bring your own power — and copyright is the half of that package where the money is.


Morphotonics, a Dutch company that has spent 12 years building nanoimprint lithography for display optics, raised 40 million euros to scale production and push into optical components for data centres. The round was backed by 3M Ventures, Innovation Industries, BOM, Invest-NL, the European Innovation Council Fund and the European Investment Bank. Its stamps pattern glass and film into waveguides — the light-guiding layer inside Meta's Ray-Ban Display and Magic Leap headsets — and a next-generation machine now in build is rated at more than 6 million waveguides a year, shipping early next year. The data-centre half is the interesting one: co-packaged optics moves traffic between servers as light instead of copper, and that interconnect is where AI clusters hit power and reach limits. Morphotonics has not shipped a machine into that market yet and says only that customers have validated the technology — so read this as a supply-chain signal, not a product.

What to watch: whether Anthropic's submission is published with the same level of detail as OpenAI's, and whether the government's final model lands closer to a licence with a payment rail — which is what both labs are now asking for, one explicitly and one implicitly.

If a country can be asked to rewrite its copyright law as a condition of investment, what should it be asking for in return? Tell us in the comments.

Read more

Akhetonics says its all-optical CPU reaches a customer in 2026

Akhetonics says its all-optical CPU reaches a customer in 2026

Light-based computing keeps promising more than it delivers — but one Munich startup has just put a date on its bet, and the interview laying it out is doing the rounds on Hacker News this week. Akhetonics says it will deploy its first commercial machine with a major customer by the end of 2026, with several more planned for 2027. The company, founded by Michael Kissner and Leonardo Del Bino, is building a computer where data enters as light, is switched as light, and circulates through memory

The Week in AI — October 5–11, 2026

The Week in AI — October 5–11, 2026

Every big claim this week turned out to rest on fine print more interesting than the headline: revenue only the company reporting it can define, safety tests sandboxed while the product keeps the web, and a Pentagon phase-out nobody would confirm until reporters kept asking. The week's top 5 1. OpenAI's revenue was $20 billion below the numbers everyone quoted — and the gap was definitional. The Financial Times reported Thursday that OpenAI's annualized revenue runs roughly $20 billion unde

Drone strike shuts a third Yandex data center, taking YandexGPT offline

Drone strike shuts a third Yandex data center, taking YandexGPT offline

Russia's largest tech company is learning what the AI era's infrastructure war looks like from the receiving end — three data centers in four days, and with them much of the cloud layer Russian businesses run on. A Ukrainian drone strike knocked out Yandex's data center in Vladimir early Sunday morning, the third of the company's facilities hit since October 8. The site — reported at roughly 50 MW and designed for about 2,880 server racks — stopped operating completely after the attack, Yandex

Agent teams cost up to 5x more, barely score higher

Agent teams cost up to 5x more, barely score higher

The multi-agent hype train hit a benchmark this weekend — and the grid and the trucking regulators had quiet weeks of their own. Vals AI put agent teams head-to-head with single agents on its Vibe Code Bench, and the teams cost between 1.8 and 5.1 times more for almost no extra quality. The evals company ran GPT-6 Sol and Claude Opus 5.5 solo and in teams across 50 apps at two reasoning efforts; out of four comparisons, only one was statistically significant — Sol at medium effort, where the t