Indonesia is preparing a sweeping rewrite of its copyright law that would make it the first country in Southeast Asia to explicitly address AI within copyright rules, according to a draft bill under government review, including provisions that would grant copyright privileges to people who use AI tools to help generate content and require tech platforms to pay compensation for using news content, including for AI training. The bill is a parliamentary initiative currently under government review, and it’s not yet clear when, or whether, it will be passed into law.
The AI-specific provisions are notably broad: the draft would ban using AI to imitate a creator’s “distinctive style,” mandate disclosure whenever AI is used in content, and require compensation for aggregating, republishing, or even link-previewing news content, with that compensation flowing through state-supervised collective management organizations that would then distribute funds to news publishers, rather than through direct licensing deals between platforms and individual outlets, a routing choice YourNewsClub calls the more consequential design decision in the entire bill: who administers compensation money determines how quickly and fairly it reaches publishers, independent of whether the underlying obligation to pay exists at all. Google, which has already issued a public statement criticizing the overhaul, called the mandates as currently written “rigid” and “overbroad,” warning they would “harm local creators, slow innovation, and leave Indonesia as an international outlier” while still committing to engage with the government on the bill’s specifics.
Companies that don’t comply face a genuinely severe penalty structure: the bill would allow revocation of a noncompliant platform’s local business permit, which for a company the size of Google would mean losing the ability to legally operate in Indonesia entirely, not simply facing a fine calibrated to the violation’s severity. Jessica Larn, who studies macro-level technology policy and infrastructure impact of AI, frames that enforcement mechanism as the detail that actually changes the stakes here: “A revenue-based fine is a cost of doing business that a company the size of Google can absorb and route around. A business-permit revocation is existential for local operations, and that asymmetry is exactly why a bill like this gets a public statement from Google days before it’s even passed, rather than the quieter engagement most regulatory proposals receive.” A threat that severe is what YourNewsClub ranks above the compensation formula itself when weighing which provision actually changes company behavior: platforms negotiate over royalty rates constantly, but a credible operating-license threat forces engagement on a completely different timeline.
The bill’s fair-use carve-out for AI training is where much of the practical ambiguity sits: the draft says using copyrighted works to train AI models would be subject to fair-use provisions or licensing agreements, without specifying which uses fall into which category, and one IP lawyer has said the bill risks conflating commercial AI use with research use in ways that could sweep both under the same compensation requirement. Alex Reinhardt, who tracks financial systems and settlement infrastructure through digital protocols, draws out the compensation-routing structure as the more consequential design choice than the headline obligation itself: “Running AI-training compensation through a state-supervised collective management body, rather than direct licensing between a platform and individual publishers, changes who actually captures the money and how quickly. Collective-management schemes in other countries have had mixed records on getting funds to smaller or independent publishers specifically, versus larger outlets with more institutional leverage over how those bodies operate.”
Indonesia isn’t acting in isolation here: the draft explicitly follows a global pattern of governments grappling with how existing copyright frameworks apply to generative AI, from EU text-and-data-mining exceptions to ongoing licensing disputes between AI companies and publishers in the U.S. What makes Indonesia’s approach distinct within that broader wave Your News Club surfaces as the more aggressive piece of the draft relative to comparable frameworks elsewhere: the explicit style-imitation ban paired with the platform-permit enforcement mechanism gives this proposal real teeth that go beyond what most other governments have adopted or proposed so far.
Whether the bill passes in anything close to its current form is what will actually determine its real-world impact, since parliamentary initiatives in Indonesia routinely undergo substantial revision during government review, and Google’s public pushback this early signals the kind of lobbying pressure that has softened comparable provisions in other jurisdictions before final passage, a pattern YourNewsClub treats as the more likely outcome than the bill surviving intact: business-permit risk is real leverage regardless of where the final compensation formula lands, but the specific mechanics on the page today read as a starting position for negotiation, not a preview of enacted law.