Skip to content

Australia Online Safety Act: BOSE and Deepfake Penalties

3 min readMahdi Kazempour, Founder, Original Pictures

Line-screen illustration of a hand holding a closed padlockLine-screen illustration of a hand holding a closed padlock
AI-generated illustration
On this page

Definition: Australia's Online Safety Act, the Basic Online Safety Expectations (BOSE), and the Criminal Code Amendment of 2024 form the regime through which the eSafety Commissioner addresses deepfake and image-based abuse.

TL;DR: The BOSE Determination (F2024L00590, 31 May 2024) sets expectations for online services, and the Criminal Code Amendment Act 2024 No. 78 created offences for sharing non-consensual deepfake sexual material, with up to 7 years for aggravated cases. Providers can face civil penalties up to A$49.5M. The eSafety Commissioner enforces.

The two halves of the regime

BOSE sets out what the government expects services to do, from minimizing harmful material to responding to the eSafety Commissioner. The Criminal Code Amendment Act 2024 then criminalizes sharing non-consensual deepfake sexual material, with aggravated offences carrying up to seven years. Together they cover both platform expectations and individual conduct.

Real enforcement

This is not theoretical. In eSafety v. Anthony Rotondo, Justice Erin Longbottom ordered a A$343,500 penalty on 26 September 2025, an early signal that the Commissioner will pursue penalties for deepfake-related breaches.

How provenance supports compliance

Verifying inbound media and signing first-party content gives a service evidence for its work on BOSE expectations and a faster path to act on non-consensual deepfakes. The audit trail matters when the Commissioner asks what the service did and when.

The incident behind this

eSafety v. Anthony Rotondo: the Federal Court, per Justice Erin Longbottom, ordered a A$343,500 penalty on 26 September 2025, demonstrating active enforcement of Australia's deepfake provisions.

Regulatory mapping

RegimeEffectiveBiteWhy it applies
BOSE Determination F2024L0059031 May 2024Up to A$49.5M civil (providers)Basic Online Safety Expectations
Criminal Code Amendment Act 2024 No. 783 Sep 2024Up to 7 years (aggravated)Non-consensual deepfake sexual material

FAQ

Who enforces the Online Safety Act?

The eSafety Commissioner, who can issue notices, seek civil penalties, and refer criminal matters, as the Rotondo case shows.

Where Original Pictures stands today

Original Pictures offers a Sign API, a Verify API, hosted verify pages and the <op-verify> embed, the AI disclosure SDK with a tamper-evident evidence log, workflow templates, and the SDKs that wrap the APIs (the embed, the disclosure SDK and the SDKs are not yet published to npm or PyPI). One POST /v1/sign with "watermark": true, "fingerprint": true attaches a C2PA-format manifest and an invisible watermark and keeps a content fingerprint for lookup. RFC 3161 timestamping is built and will be switched on with our production signing certificate; files signed before then carry no trusted timestamp. Any C2PA-aware validator can read the manifest without calling us.

Original Pictures is C2PA conformant: it is listed on the C2PA Conforming Products List as a generator product under the C2PA Conformance Program 0.2 (C2PA Technical Specification 2.4), and any C2PA-aware validator can read its manifests. Signed with our production certificate from SSL.com, on the C2PA Trust List; validators that use the C2PA Trust List show the signer as recognized. One thing is on the near roadmap, and we name it as roadmap, not as shipped: a consumer capture app (not yet available). We do not sell a capture SDK.

Bottom line: Australia's regime is active and penalties are real. Verify inbound media, sign first-party content, and keep an audit trail that supports your work on BOSE expectations and fast action on deepfakes.

Try it on your own files

Create an API key and sign a test file. Talk to us if you need a walkthrough.