Singapore Criminalises AI-Generated Intimate Images as Law's Second Phase Kicks In

Singapore has spent the past year racing to make itself an AI hub — and now it is writing the rules for the technology's darkest edge cases. From Monday (Aug 17), producing intimate images without consent — including fully synthetic, AI-generated ones — becomes a specific criminal offence in Singapore, closing a legal gap that has nagged lawmakers since generative AI made it trivial to fabricate convincing imagery of anyone.

The change arrives with the second phase of the Criminal Law (Miscellaneous Amendments) Act 2025, passed by Parliament in November. The definition of "intimate image" under the Penal Code now explicitly covers synthetic AI-generated material — images generated to depict a person without altering an original image or recording. Producing such material without consent carries up to two years' jail, a fine, or both, and if the victim is under 14, imprisonment is mandatory with the offender also liable to a fine or caning. The same package clarifies that child abuse material includes computer-generated content even without proof that a real child's image was used, and extends sexual grooming offences to overseas meetings when either party's travel began in Singapore — with penalties for victims under 14 rising from four to seven years.

This is the enforcement layer of a broader policy push that has been building for a year. When the amendments were tabled in October 2025, the Ministry of Home Affairs argued that "completely synthetic images" generated by increasingly sophisticated AI models are "no less objectionable" than images made by altering real ones. Home Affairs Minister Sim Ann framed the intent as both "to reduce the roadblocks to enforcement and to reduce the supply of such materials, which contributes to a vicious cycle of abuse." The update sits alongside Singapore's other AI guardrails — the Model AI Governance Framework, the chatbot transparency code, and the AI-specific privacy guidelines issued last month — where the consistent thread is that synthetic content should be treated with the same seriousness as physical-world harms.

The law also modernises adjacent areas: enhanced penalties now cover the large-scale electronic circulation of obscene material to 10 or more people (up to two years, more when minors are depicted), a new offence targets those who set up or manage online locations — think Telegram groups and websites — to facilitate such circulation, and caning becomes discretionary for crimes like attempted robbery and carrying an illegal weapon. For AI developers and platforms, the signal is unambiguous: image-generation tools must be designed with consent in mind, and the era of treating deepfake production as a grey area is over in Singapore.

Why it matters for Singapore: this makes Singapore one of the first jurisdictions in Asia to extend criminal liability explicitly to fully synthetic AI imagery, in a year when the city is courting the very companies building these models. It is a deliberate balancing act — welcoming AI investment while drawing clear lines on abuse — and it gives the ecosystem something concrete to point to when regulators elsewhere ask how to handle AI-generated harms. Watch for follow-on platform guidance and early test cases as the provision meets its first prosecutions.