CANBERRA, Australia — Australia’s government is preparing legislation that would place a proactive safety duty on online services and give the eSafety Commissioner stronger tools to examine how platforms make decisions.
Communications Minister Anika Wells outlined additional elements on 7 September, including greater user control over algorithmic recommendations and powers allowing the regulator to question company executives and engineers. The final bill must still pass parliament, and details may change.
The government’s May framework already makes the gender stakes explicit. It says services offering artificial-intelligence tools should take measures to prevent users generating or sharing illegal or seriously harmful material, including images that “nudify” a person without consent. Existing complaint schemes for image-based abuse would remain.
That is an important change in regulatory logic. A takedown process begins after harm occurs and often after an intimate image has been copied across services. A duty of care asks whether a company tested foreseeable misuse before launch, changed risky features and measured whether its safeguards work.
The proposed framework would cover social networks, messaging and dating services, search engines, app stores, chatbots and image generators. Higher-risk services could face regular assessments, transparency requirements, audits and penalties. Government rules would be subject to parliamentary disallowance and human-rights statements.
Civil-liberties concerns deserve serious treatment. Ministers should not receive vague authority to define harmful speech without clear thresholds, independent review and appeal. The law must distinguish unlawful abuse and engineered amplification from controversial but lawful expression.
For women and girls, enforcement should test outcomes rather than policy pages. Regulators need data on how quickly non-consensual images are blocked, whether repeat uploads are detected, how non-users can report abuse and whether safety tools work across languages and skin tones. Researchers and survivor organisations should have protected access to scrutinise claims.
The bill also proposes user choice over recommendation systems. An opt-out can reduce unwanted exposure, but defaults are powerful; legislators should examine whether high-risk features should require affirmative consent instead.
Australia is not yet offering a completed solution. It is proposing a framework that recognises product design as part of the harm. If the final law preserves due process while demanding evidence from platforms, women will no longer have to prove only that one image injured them. Companies will have to show why their systems made that injury so easy.
Join the discussion