.png)
SHAH ALAM — What happens when technology can create an image of a sexually abused child without a camera ever being pointed at a real child?
For a US federal appeals court, the answer has produced a deeply unsettling legal paradox.
The court has upheld the dismissal of a possession charge against a Wisconsin man accused of creating and possessing AI-generated child sexual abuse material (CSAM), saying existing US Supreme Court precedent leaves it constitutionally unable to criminalise the private possession of certain entirely synthetic images.
The ruling does not mean all AI-generated CSAM is legal in the United States. Nor does it protect the production or distribution of such material.
But it establishes a troubling boundary: where an image depicts a fictional child, no real child was involved in its creation and the material is privately possessed, decades-old First Amendment precedent can prevent federal prosecution on the possession charge.
That is what makes the case so disturbing.
The technology has changed. The law has not.
The case concerns Steven Anderegg, a Wisconsin man arrested in 2024 after investigators allegedly found thousands of AI-generated sexually explicit images depicting minors.
US prosecutors charged him with producing, distributing and possessing the material, as well as transferring similar material to a minor under 16.
The lower federal court dismissed the possession charge. The government appealed.
On Aug 25, the US Court of Appeals for the Seventh Circuit upheld that dismissal.
The case was decided by a three-judge panel comprising Judges John Z. Lee, Robert L. Pryor and Joshua Kolar, with Lee writing the opinion.
The other charges remain. That distinction matters.
The court did not declare a general right to create, distribute or send AI-generated CSAM to children. But on the specific possession charge before it, the judges concluded that Supreme Court precedent tied their hands.
They were not oblivious to the implications.
At the heart of the case is the 2002 Supreme Court decision Ashcroft v Free Speech Coalition.
The Supreme Court struck down parts of a federal law criminalising certain sexually explicit images depicting fictional children where no actual child was involved in their production.
The reasoning rested partly on a constitutional distinction: the government's compelling interest in protecting children from sexual exploitation is strongest when an actual child has been exploited.
That distinction was formulated in an era when computer-generated images were crude and obviously artificial.
Generative AI has shattered that assumption.
Today's systems can create extraordinarily realistic images, including depictions of child sexual abuse that may be virtually indistinguishable from photographs of actual abuse.
Judge Lee acknowledged the problem.
But he said the Seventh Circuit was bound by Supreme Court precedent. The judges had concerns about the boundaries created by those cases, he wrote, but were “not free to redraw them ourselves.”
That may be the most consequential sentence in the ruling.
The technology has moved. The constitutional framework has not.
Supporters of the legal distinction may argue that an entirely synthetic image has no physical child victim.
That is precisely where child-protection advocates should be cautious.
The absence of a camera, physical assault or identifiable child during production does not automatically make the material harmless.
Researchers have warned that AI-generated CSAM can facilitate grooming, coercion and sexual extortion, contribute to the normalisation of sexualised images of children and lower barriers to offending.
It can also be used to manipulate or recreate images of real children.
And there is a further problem: investigators increasingly have to determine whether apparently realistic material is synthetic or whether an actual child was abused.
That distinction is not academic.
It can determine whether there is a real child somewhere who needs to be identified, rescued and protected.
US prosecutors have warned that the growing volume of AI-generated material could burden investigators and make it harder to identify genuine victims.
The more uncomfortable question is what happens next.
If increasingly realistic synthetic CSAM becomes cheap, accessible and easy to produce, society could face a vast volume of material that looks like child abuse even when investigators cannot immediately determine whether an actual child was involved.
That presents an extraordinary challenge for police, technology companies, prosecutors and courts.
It also creates opportunities for predators.
AI can potentially produce thousands of images without a studio, camera, accomplice or physical victim, dramatically lowering the practical barriers to creating abusive material.
That does not prove AI-generated CSAM causes someone to become a paedophile.
But it does mean society must take seriously the possibility that easy access to increasingly realistic sexualised depictions of children could normalise, reinforce or facilitate abusive interests.
The International Watch Foundation said it assessed 8,029 AI-generated images and videos in 2025 as depicting realistic child sexual abuse.
UNICEF has called for countries to criminalise AI-generated content depicting child sexual abuse, warning that children cannot wait for laws to catch up with technology.
For Malaysia, the case should be a warning rather than a distant American legal curiosity.
The country is already confronting the misuse of generative AI to produce sexualised material involving women and children.
In January, Malaysian authorities moved against X over concerns about the misuse of its Grok AI system to create sexualised and manipulated images, including material involving minors.
Malaysia has also recognised that existing laws may need to evolve.
Digital Minister Gobind Singh Deo said in June that the government was moving to strengthen legislation addressing AI-enabled offences, including deepfakes and AI-generated child sexual exploitation material.
The Communications Ministry has separately said AI-generated CSAM can potentially be prosecuted under existing Malaysian laws.
That makes the American case particularly relevant.
Malaysia should not wait until a court is forced to confront the same technological contradiction.
The question should be settled before the loophole becomes an established defence.
The world is entering a period in which the distinction between real and fake is becoming increasingly difficult to sustain.
A synthetic image may depict a fictional child. But the technology that created it is real, the person consuming it is real, the platform distributing it is real and the predator using it is real.
So are the potential consequences for children.
That is why describing AI-generated CSAM as “fake” can be dangerously misleading.
The absence of a physical victim during creation does not place the material outside the wider ecosystem of child sexual exploitation.
That ecosystem can include grooming, extortion, manipulation and the recycling of imagery involving genuine victims.
The legal system therefore faces a fundamental question: should the law continue to treat an image differently simply because a machine, rather than a camera, produced it?
In the US, the Seventh Circuit did not answer that question. It said the Supreme Court must.
The judges have effectively acknowledged that a constitutional framework created before generative AI existed may now be struggling to address what the technology makes possible.
The next move belongs to lawmakers and the Supreme Court.
But the warning belongs to everyone.
If technology can manufacture convincing images of children being sexually abused faster than the law can respond, investigators may increasingly struggle to determine where fiction ends and a real child's suffering begins.
Children should not have to wait for the law to catch up.
