Canada Has Made Intimate Deepfakes Illegal. So Why Can Canadians Still Access Websites Designed to Create Them?
- The White Hatter
- 9 hours ago
- 21 min read
CAVEAT - Last week we received this email that was the catalyst for this article:
Upon closer scrutiny of the source and content of the email, we determined that it was likely AI generated and connected to a deepfake pornography website. The service allows users to upload or capture an image of virtually anyone’s face and use its AI tools to generate fabricated sexualized content without that person’s knowledge or consent. The platform openly markets the ability to transform ordinary photographs into short AI-generated “undress” videos and offers users a catalogue of more than 120 sexualized scenarios in which a person’s likeness can be digitally inserted.
What makes services like this particularly concerning is that the technology itself does not necessarily know whether the person depicted in the original photograph is an adult. A publicly available image can potentially be taken from social media or elsewhere online and repurposed to create fabricated intimate imagery, illustrating how easily an ordinary photograph can be weaponized through generative AI.
There is a growing category of artificial intelligence websites that we believe parents, caregivers, educators, policymakers, law enforcement, and legislators need to pay much closer attention to. These services are commonly marketed using terms such as AI “nudify,” “undress,” or “clothing removal” tools, something that we have addressed in the past (1). Although individual services operate differently, the basic concept is remarkably simple, a user uploads an ordinary photograph of a real person and artificial intelligence generates a new image that appears to depict that person partially or completely nude, or in some cases in a sexualized context.
What makes these technologies particularly concerning is that the person depicted does not need to participate in the process at all. They do not need to provide the photograph, create an account, download an app, communicate with the person generating the image, or even know that their photograph has been uploaded. Most importantly, they do not need to consent. An ordinary photograph copied from Instagram, TikTok, Facebook, a school website, a sports team page, a graduation announcement, a dating profile, or even a family photograph can potentially become the source material used to manufacture a realistic-looking intimate image.
This creates an uncomfortable but increasingly important Canadian policy question, “If Canadian law now recognizes certain realistic non-consensual intimate deepfakes within its intimate image laws, why can Canadians still readily access commercial websites specifically designed to help people create them?”
Before going further, we want to make something very clear. This article is not an argument for unrestricted government censorship of the Internet, nor are we suggesting that governments should be given broad powers to block websites simply because they find the content offensive, controversial, politically inconvenient, or socially undesirable. Such authority should concern Canadians regardless of their political beliefs. Freedom of expression, access to information, procedural fairness, and limits on government power are fundamental considerations in any discussion about Internet blocking.
However, there is also an important distinction between a website containing controversial expression and a commercial service whose central product is allowing someone to take an image of an identifiable real person and manufacture a realistic nude or sexual depiction of them without their participation or consent. We believe that distinction deserves a serious evidence based discussion rather than being dismissed as either “Internet censorship” on one side or “online safety” on the other.
First, Understand What These Websites Actually Do
The word “deepfake” can make this technology sound highly technical, perhaps suggesting that someone needs advanced knowledge of artificial intelligence, graphics software, coding, or digital manipulation to create a convincing image. That may once have been true, however, the technological complexity is hidden behind extremely simple consumer interfaces. From the user’s perspective, the process may involve little more than finding a photograph, uploading it, selecting an option, and waiting for the artificial intelligence system to generate the result. This was the case with the product referenced in the email that was sent to us.
That simplicity fundamentally changes the scale of the problem. A person who previously lacked the technical ability to manipulate a photograph can now potentially outsource that work to an automated service. This lowers the barrier to entry from technical expertise to basic Internet access. A teenager targeting a classmate, for example, may no longer need sophisticated editing skills or expensive software. They may only need a photograph, a web browser, and access to one of these services.
This becomes particularly relevant for young people because ordinary photographs are everywhere. Schools publish photographs of students participating in athletics, theatre, music, clubs, awards ceremonies, and graduation events. Sports organizations post team photographs. Parents share family pictures. Young people themselves use social platforms to communicate and document their lives. Even a young person who maintains relatively restrictive privacy settings cannot necessarily control every photograph of themselves that exists online.
The original photograph may be completely innocent and non-sexual. It may show a teenager wearing a school uniform, standing beside friends, playing soccer, attending a graduation ceremony, or posing for a family photograph. The problem is not necessarily what the young person posted. The problem is what someone else can now do with that photograph, and that represents an important shift in the online intimate image landscape.
This Is Not A Hypothetical Concern
Here at The White Hatter, we have assisted individuals dealing with AI manipulated intimate imagery. We have seen how distressing it can be when an ordinary photograph is transformed into sexualized content and then circulated as though the resulting image were authentic. For the person depicted, knowing that the image is fake does not necessarily eliminate the embarrassment, fear, reputational consequences, or feeling that control over their identity and body has been taken away from them.
For many years, online safety education surrounding intimate images focused heavily on preventing young people from taking or sending sexual photographs. Parents and educators were encouraged to speak with teens about sexting, privacy, permanence, screenshots, forwarding, and the possibility that an intimate image voluntarily shared with one person could later be distributed without permission. Those conversations remain important, particularly because consensually created intimate images can still be weaponized after relationships end or accounts are compromised.
Artificial intelligence, however, has fundamentally changed the equation because an authentic intimate photograph is no longer required. A young person can make every decision adults traditionally encouraged them to make. They can refuse to send nude photographs, avoid sexting, maintain private accounts, carefully consider what they post, and still become the subject of an apparently intimate image created by somebody else.
This means that one of the traditional messages used in intimate image prevention, “Don’t send a nude because you may lose control of it,” is no longer sufficient on its own. Increasingly, there may never have been an intimate photograph for the person to lose control of in the first place. Someone else can manufacture it.
That distinction also matters when supporting survivors of a non-consensual sent intimate image. Asking a targeted young person why they took or sent an image may be completely irrelevant in a deepfake case and can inadvertently place responsibility on the person who was targeted rather than on the person who deliberately manufactured or distributed the image.
Canadian Law Has Started Catching Up
Canada took an important step in 2026 with the passage of the “Protecting Victims Act”, S.C. 2026, c. 19, which received Royal Assent on June 18, 2026. Among its many amendments to the Criminal Code, the legislation amended section 162.1 dealing with non-consensual intimate images. The new definition of an “intimate image” captures not only conventional visual recordings but also certain visual representations made by electronic or mechanical means, including through artificial intelligence. In the case of an artificially generated representation, the legislation addresses depictions of an identifiable person as nude, nearly nude, exposing their sexual organs, or engaged in explicit sexual activity where the depiction is likely to be mistaken for an actual visual recording of that person (2).
Justice Canada’s Charter analysis of the legislation specifically identifies these types of fabricated intimate images as “deepfakes.” Importantly, the department recognizes that the harm arising from distribution of a deepfake intimate image can be similar to the harm caused when an authentic intimate image is distributed without consent. The government also acknowledges that restrictions on distributing such material engage freedom of expression under section 2(b) of the Canadian Charter of Rights and Freedoms, which is important when considering the broader constitutional issues discussed later in this article (3).
The Protecting Victims Act also strengthened the potential criminal consequences associated with non-consensual intimate images. Among other changes, the legislation increased the maximum sentence for the indictable form of the non-consensual distribution offence to 10 years imprisonment and created an offence relating to knowingly threatening to publish, distribute, transmit, sell, make available, or advertise an intimate image without the depicted person’s consent in the circumstances set out in the legislation.
These changes matter enormously because they recognize something victims have understood for some time, the harm created by a realistic fabricated sexual image does not necessarily disappear simply because investigators, parents, or the person depicted can eventually prove that artificial intelligence generated it.
Consider what this might look like inside a Canadian high school. A realistic looking nude image purporting to depict a 15 year old student begins circulating through group chats. The teenager knows immediately that it is fake. Their parents eventually learn that it is fake. Perhaps the school determines that artificial intelligence was involved, and perhaps police eventually identify how the image was created. None of that guarantees that every student who saw the image will learn the truth. Screenshots may continue circulating long after the original post disappears, and the teenager may have little ability to know who has seen, saved, or redistributed it.
The social and emotional consequences can therefore be very real even when the pixels themselves are synthetic. The humiliation can be real, the loss of privacy can feel real, the fear about who has seen the image can be real, the damage to relationships or reputation can be real. and the fact that artificial intelligence generated the depiction does not magically erase those consequences.
Canada has therefore begun adapting its criminal law to an environment where a convincing intimate image no longer needs to originate from a camera. That is important progress, but it leads directly to another question that we believe deserves considerably more attention, “What about the commercial services specifically designed to manufacture these images in the first place?”
Illegal Output, Yet The Tool Can Remain Accessible
Imagine a Canadian teenager who finds an ordinary photograph of a classmate on Instagram. They download the photograph and upload it to an overseas AI nudification service. The website processes the photograph and produces what appears to be a nude depiction of the classmate. The teenager saves the result and begins distributing it through private messages, group chats, or social media. This is not a hypothetical, but in fact a scenario that we helped a family with last year.
Depending on the specific facts, Canadian law may provide mechanisms for investigating what happened and potentially prosecuting the person responsible for certain conduct if the elements of an offence are established. There may also be mechanisms available to attempt to remove the resulting image, identify where it has been distributed, preserve evidence, or involve law enforcement and child-protection resources. As an example, here in British Columbia we have a free resource, the “Intimate Image Protection Service” (4).
However, after all of that, the website used to manufacture the image may remain readily accessible from Canada. Another student can potentially visit it the following day, upload another photograph, and repeat the process. The service itself may be hosted outside Canada, in places like Russia, operated by individuals who are difficult to identify, and designed specifically to profit from making this type of image generation easy.
This is where we believe there is an important policy discussion that Canada has not yet fully addressed. We have begun developing laws aimed at what happens once certain intimate deepfakes exist and are distributed, but we also need to ask what responsibility should attach to commercial infrastructure deliberately built around making their creation simple and scalable. That does not automatically mean these websites should be blocked, it means a question deserves to be asked, “Could Canada actually block one of these websites?
From a purely technical perspective, “yes, they could”. Canadian Internet providers have the ability to interfere with access to specified domains, IP addresses, URLs, or other Internet resources using different technological methods. The much harder questions are whether they should be permitted or required to do so, who should have the authority to make that decision, what evidence should be required, and what safeguards should exist to prevent abuse and over blocking.
Canada does not operate a general national Internet firewall where the federal government can simply place an unwanted website on a blacklist and make it disappear for Canadian users, and that limitation is important. Internet service providers should not become private censorship authorities, nor should governments have unrestricted power to decide what Canadians are allowed to read, watch, discuss, or access online.
Targeted website blocking, however, is not completely foreign to Canadian law. In Teksavvy Solutions Inc. v. Bell Media Inc., 2021 FCA 100, the Federal Court of Appeal upheld Canada’s first site blocking order in a copyright case. Canadian Internet service providers that were not themselves accused of copyright infringement were required to prevent subscribers from accessing specified services associated with infringement. The Supreme Court of Canada subsequently declined to hear a further appeal (5)(6).
We need to be careful about what this precedent does and does not establish. A copyright site blocking injunction does not mean a court can automatically take the same legal reasoning and apply it to an AI nudification service. The legal rights being protected, statutory authority, parties involved, evidentiary record, proportionality analysis, and potentially applicable Charter considerations could all be different.
What the case does demonstrate, however, is that requiring Canadian Internet providers to prevent customers from accessing specified Internet resources is not technically or legally unimaginable in Canada. Under appropriate circumstances and legal authority, Canadian courts have already ordered it.
The policy question therefore becomes whether Parliament should establish an appropriately narrow legal mechanism that could potentially apply when compelling evidence demonstrates that a commercial service is principally designed to facilitate the creation of realistic non-consensual intimate imagery, and we believe this is something that should be explored here in Canada.
Canada Already Has Experience Blocking Certain Foreign Internet Resources
There is another Canadian example that deserves attention because it involves something considerably more serious than copyright infringement. Cleanfeed Canada was launched in 2006 through the Canadian Coalition Against Internet Child Exploitation. According to Cybertip.ca, its purpose was to block customer access to certain non-Canadian websites hosting child sexual abuse material. Cybertip.ca securely supplied the relevant list to participating Canadian Internet service providers, which participated voluntarily (7).
The initiative subsequently evolved. The Canadian Centre for Child Protection reports that the Cleanfeed Canada URL list was transitioned into the Shield API by Project Arachnid, providing a more comprehensive list of URLs to a wider range of electronic service providers. Cybertip.ca also states that Cleanfeed continues to be used by Canadian ISPs today as part of Shield.
Project Arachnid itself takes a broader approach to reducing the availability of child sexual abuse material. Developed by the Canadian Centre for Child Protection and launched in 2017, the platform crawls the open web searching for CSAM. When CSAM or other harmful abusive content is detected and classified, removal requests can be sent to hosting providers. According to Cybertip.ca, the system is capable of processing thousands of images per second, allowing harmful material to be identified at a scale that would be extremely difficult through traditional manual processes alone (8).
This is relevant because it demonstrates that the concept of preventing Canadians from accessing specified foreign Internet resources is not theoretical. Canada has experience with systems designed to identify, classify, remove, and in certain circumstances restrict access to particularly harmful material.
That does not mean the same framework should automatically be expanded to AI nudification services. Child sexual abuse material occupies a particularly serious and legally distinct category, and any attempt to extend blocking mechanisms beyond that context would require careful legal and constitutional consideration.
However, it does demonstrate something important for the broader policy discussion, the technical capability already exists. The harder issue is establishing the legal threshold for when using it is justified.
When Children Can Be Uploaded, The Question Becomes Even More Urgent
Imagine a commercial website advertising that users can upload virtually any photograph and generate a nude or sexualized representation. How does that service reliably know whether the person depicted in the uploaded photograph is 25 years old, 17 years old, or 14 years old? What happens when the photograph came from a Grade 8 sports page, a teenager’s social media account, a school yearbook photograph, or a parent’s family account?
If a commercial service is built to accept photographs of real people and generate sexualized depictions, the possibility that photographs of minors will be uploaded should not be treated as an obscure or unimaginable misuse. Young people’s photographs are widely available online, and the very simplicity of these tools lowers the barrier for other young people, or adults, to experiment with them.
Depending upon precisely what an artificially generated image depicts and the circumstances surrounding its creation, possession, access, or distribution, other provisions of Canadian criminal law concerning child sexual abuse material may also require consideration. Careful legal analysis is important here because we should not automatically describe every AI generated sexualized depiction involving a minor as satisfying every element of Canada’s criminal definition of child sexual abuse material, the depiction and surrounding facts matter.
However, the broader policy question remains legitimate, “Why should intervention necessarily begin only after someone has generated the sexualized image of a child?”
If a service’s predominant commercial purpose is to take photographs of identifiable people and manufacture realistic nude representations, policymakers should be asking what safeguards that company has implemented to prevent children from being depicted, how those safeguards are tested, whether they are effective, and what happens when abuse is identified.
Not Every AI Image Generator Should Be Treated The Same
One of the most important distinctions in this discussion is between a general purpose technology that can be misused and a specialized service whose central function is the activity causing concern.
A general purpose AI image generator might theoretically be manipulated into producing inappropriate content despite safeguards designed to prevent that outcome. The same principle applies to many technologies. Cameras can be misused, photo editing software can be misused, and messaging applications can be misused. The fact that a technology is capable of misuse is not, on its own, sufficient justification for banning or blocking it.
A service whose principal marketing proposition is essentially “upload someone’s photograph and see them naked” presents a substantially different policy question. In that situation, the controversial use is not necessarily incidental to the product, it may be the product.
Any Canadian legislative response should therefore be carefully constructed around factors such as the predominant function of the service, how it is marketed, whether it encourages users to upload images of identifiable third parties, what safeguards exist, whether meaningful consent mechanisms are present, how effectively minors are protected, and what the service does when abuse is reported.
These distinctions are essential because overly broad legislation could inadvertently capture legitimate artificial intelligence tools, image editing services, artistic applications, health technologies, or other platforms that have lawful purposes.
The goal should not be to regulate “AI that can make nudity.” The policy discussion should be about services deliberately designed and commercially promoted to facilitate the creation of non-consensual sexualized representations of identifiable people.
Why Can’t Canadian Internet Providers Simply Block Them Today?
We are not lawyers, however, in our research for this article here’s what we learned.
Canadian telecommunications law makes this considerably more complicated than simply identifying an offensive website and telling Internet providers to make it disappear.
Section 36 of Canada’s Telecommunications Act generally provides that a Canadian carrier cannot control the content or influence the meaning or purpose of telecommunications it carries for the public without CRTC approval. That principle serves an important purpose because Canadians should not have Internet providers independently deciding which lawful websites, political perspectives, news organizations, advocacy groups, social movements, or controversial opinions their customers are permitted to access.
The CRTC has historically treated Internet blocking cautiously. In 2018, the FairPlay Coalition asked the Commission to establish a regime through which websites overwhelmingly or structurally engaged in copyright piracy could be identified and Canadian ISPs required to block them. The CRTC rejected the application because it concluded that it did not have jurisdiction under the Telecommunications Act to implement the proposed mandatory copyright blocking regime. The Commission’s decision also discusses concerns surrounding censorship, net neutrality, over blocking, under blocking, effectiveness, circumvention, and judicial oversight (9).
An even more relevant development occurred on June 18, 2026, when the CRTC expanded its network level blocking framework. The revised framework permits participating Canadian carriers to use approved network level blocking methods to address malicious Internet traffic associated with cybersecurity threats, subject to principles that include necessity, accuracy, and consumer privacy. The framework also requires mechanisms for dealing with false positives and overblocking (10).
However, there is a particularly important limitation in the CRTC’s decision that directly relates to this article. The Commission states that the authorization applies to blocking for cybersecurity purposes and does not authorize blocking for other purposes, including blocking activity merely because it is otherwise illegal, or blocking for commercial, competitive, or political purposes. That distinction is extremely important.
It means we should not assume that because conduct facilitated through a website may constitute a criminal offence, Canadian Internet providers can simply decide, “This website facilitates illegal intimate deepfakes, therefore we’re blocking it.”
It would appear to us, whether conduct occurring through a website may violate Canadian criminal law and whether an Internet carrier has lawful authority to block access to the entire service are two separate legal questions.
This is precisely where we believe there may be a policy gap worth examining. Canada’s current CRTC cybersecurity blocking framework is not a general purpose illegal content blocking regime. If Parliament believes a very narrow category of commercial services principally designed to facilitate image based sexual abuse should potentially be subject to blocking, a specific legal framework with appropriate safeguards may be required rather than attempting to stretch existing cybersecurity powers beyond their intended purpose.
Website Blocking Is Only One Part Of The Puzzle
Even if Canada developed a legal mechanism for blocking certain services, we do not believe website blocking should be viewed as the sole or even necessarily the most effective intervention.
Online commercial services depend on an ecosystem of infrastructure. They may need domain names, hosting, servers, payment processing, advertising, search visibility, social media promotion, analytics, email services, and sometimes mobile applications. Each of those relationships can potentially become an intervention point when a service violates applicable laws or provider policies.
Hosting companies may be able to remove services that violate their terms or applicable law. Domain registrars and registries may have abuse policies. Search engines may be able to de-index certain services under appropriate circumstances. Payment processors can determine that particular business models violate their acceptable use requirements. Advertising networks can refuse to monetize them. App stores can remove associated applications. Law enforcement can investigate Canadian operators or users and seek international cooperation when relevant offences cross borders.
This layered approach matters because blocking a single domain may accomplish very little if an operator can register another domain the next morning and continue operating in an ongoing game of digital whack-a-mole. In some circumstances, disrupting payment processing, hosting, advertising, or other infrastructure could create considerably greater friction than blocking one web address.
The goal should therefore not simply be to make a domain disappear from Canadian browsers. The broader objective should be reducing the ability of commercial enterprises to profit from facilitating non-consensual sexual manipulation while preserving lawful technologies and legitimate expression.
Perhaps Canada Needs A New, Very Narrow Legal Mechanism
We believe Parliament should at least examine whether Canadian law needs a narrowly defined mechanism specifically addressing commercial services whose predominant purpose is the creation of realistic non-consensual sexual deepfakes of identifiable people. However, any such authority would require significant safeguards precisely because website blocking is such a powerful intervention.
We would not support giving a government minister, police officer, regulator, or Internet service provider unrestricted authority to secretly decide which websites Canadians can access. A defensible Canadian framework would likely require clearly defined statutory criteria, credible evidence concerning the service’s predominant purpose and operation, independent or judicial authorization, narrowly tailored orders, transparency requirements, periodic review, mechanisms for correcting erroneous blocking, and meaningful opportunities to challenge a decision.
A court considering such an application should also examine proportionality. Blocking an entire general purpose AI service because a small number of users found a way to circumvent its safety systems would be very different from considering action against a commercial platform whose advertising, interface, and business model revolve around generating nude representations of identifiable third parties.
This is also why independent oversight matters. Decisions affecting Internet access should not be driven by political pressure, moral panic, media headlines, or a single complaint. Evidence should matter, purpose should matter, scale should matter, Safeguards should matter, and alternatives should matter.
Canada has already wrestled with many of these questions in previous website blocking debates. The CRTC’s 2018 FairPlay proceeding considered issues including judicial oversight, transparency, circumvention, overblocking, net neutrality, effectiveness, and freedom of expression. We therefore would not be beginning this policy discussion from scratch.
There is a Charter issue here too that we need to be alive to. Any serious proposal to block websites in Canada must acknowledge section 2(b) of the Canadian Charter of Rights and Freedoms and its protection of freedom of expression. Internet access has become deeply intertwined with how Canadians communicate, obtain information, participate in political and social discussions, conduct business, and express themselves. Government interference with access to online material therefore deserves careful constitutional scrutiny.
Justice Canada’s own Charter analysis of the intimate image deepfake amendments recognizes that restricting the distribution of expressive material engages section 2(b), while explaining the considerations the government believes support the amendments’ consistency with the Charter. Significantly, Justice Canada points to the similarity between the harms caused by distributing authentic intimate images and realistic deepfake intimate images.
These constitutional protections should not be treated as an inconvenience simply because the technology being discussed is disturbing. In fact, strong procedural safeguards become particularly important when governments seek authority over objectionable or harmful online material because legal powers created for the worst cases can sometimes expand into less clear circumstances over time.
For us, therefore, the question is not, “Should Canada be able to censor websites we don’t like?” The narrower and more useful question is. “Should a Canadian court, operating under clearly defined legislation and meaningful procedural safeguards, have the authority to restrict access to a commercial service when compelling evidence demonstrates that its predominant purpose is facilitating image based sexual abuse?” Those are fundamentally different propositions.
Blocking Is Not A Magic Solution
Even if Canada enacted legislation tomorrow allowing carefully targeted blocking of certain nudification services, it would be a mistake to tell parents, caregivers, or policymakers that the problem had been solved.
Domains can change, users can employ VPNs, alternative DNS services can circumvent some blocking methods, new websites can appear quickly, and AI models can potentially be downloaded and operated locally (11). Similar functionality can migrate into private communities, messaging services, or applications. Technology can evolve much faster than legislation.
The CRTC itself recognized in its 2018 FairPlay decision that different website-blocking methods involve different trade-offs concerning cost, effectiveness, over blocking, under blocking, and ease of circumvention. Its 2026 cybersecurity framework similarly requires participating carriers to minimize false positives and over blocking and to maintain processes through which customers can challenge erroneous blocking.
Website blocking should therefore be understood as a “risk reduction” measure, not a “risk elimination” measure. If a block prevents some young people from casually finding and using a nudification website, that may still have value. Friction matters, and making harmful behaviour more difficult can reduce impulsive misuse. But we should not confuse reducing accessibility with eliminating capability.
This reflects something we repeatedly emphasize at The White Hatter. Regulation has a role, platform design has a role, law enforcement has a role, schools have a role, technology companies have a role, parents and caregivers have a role, and digital literacy has a role. None of them can solve the problem alone.
Parents Also Need To Update The Conversation At Home
For parents and caregivers, perhaps one of the most important lessons from AI generated intimate imagery is that our traditional conversations about sexting and intimate photographs need to evolve, it’s no longer enough to tell young people, “Don’t send nude pictures.” Parents should certainly continue discussing the risks associated with voluntarily creating and sharing intimate images, but young people also need to know what to do when somebody creates an image they never took.
A child should understand that if someone manufactures a sexualized image of them, they can come to a trusted adult without immediately being blamed or interrogated about what they did wrong. They need to understand that being depicted in a fabricated image does not mean they participated in creating it. They should know that if someone threatens to distribute the image unless money, additional photographs, or sexual content is provided, paying or negotiating may escalate the situation rather than end it (12).
Young people should also understand the importance of preserving relevant evidence where safe and appropriate, reporting harmful content through available mechanisms, and involving trusted adults, schools, law enforcement, or specialized support organizations when circumstances warrant it.
However, prevention cannot focus exclusively on potential victims. Young people also need to hear a very clear message about their own behaviour, and that creating a fake nude of another person is not a harmless prank simply because artificial intelligence generated it. In fact, here in Canada it is now illegal!
Sometimes technology creates psychological distance from the human consequences of an action. A teenager may see the process as clicking a button on a funny website rather than creating sexualized content depicting a real classmate. That is exactly where education about consent, empathy, dignity, and consequences becomes essential.
AI generated the pixels, but a human still made the decision to upload the photograph. A human still chose the person being targeted. A human may still decide to save, show, threaten with, or distribute the result. Technology does not erase human responsibility.
Technology Has Changed. Our Thinking Must Change With It.
For years, online intimate image prevention focused primarily on controlling the original photograph, don’t take it, don’t send it, don’t share it, and understand that once an intimate photograph leaves your device you may lose control over where it goes. Artificial intelligence has changed that equation because today someone else can manufacture the photograph.
That means prevention can no longer rest primarily on telling potential victims to make better choices with their intimate images. We also need to address the choices of people creating fabricated images, the behaviour of people distributing them, the companies profiting from their creation, and the technological infrastructure that can make producing them fast, inexpensive, accessible, and scalable.
Canada has begun modernizing its criminal law to recognize this new reality. Expanding intimate image protections to include certain AI generated representations is an important development, particularly because it acknowledges that fabricated sexual imagery can produce genuine human harm.
However, legislation dealing with what happens after an image exists should not prevent us from asking what responsibilities should exist before that image is generated. If a commercial website’s predominant purpose is allowing users to take photographs of real, identifiable individuals and manufacture realistic nude or sexual depictions of them without consent, Canadians should at least be willing to discuss whether unrestricted access to that service remains appropriate.
That discussion needs nuance, it needs evidence, constitutional safeguards, judicial oversight, technical expertise, transparency, and it needs to distinguish between general purpose technologies that can be misused and commercial services specifically designed around non-consensual sexual manipulation. Most importantly, this should not become another excuse for fear based Internet regulation. At The White Hatter, our position remains Facts Not Fear.
We should not regulate artificial intelligence simply because AI frightens us, nor should every emerging technological risk become justification for another broad Internet ban. At the same time, we should not ignore technologies deliberately designed to make violating another person’s sexual privacy faster, cheaper, easier, and scalable simply because the website providing the technology happens to operate outside Canada.
There is considerable space between doing nothing and creating a government controlled Internet firewall. Perhaps it is time for Canada to seriously examine what that middle ground should look like.
Call To Action: If you are a Canadian reading this article, please forward it to your member of parliament
Digital Food For Thought
The White Hatter
Facts Not fear, Facts Not Emotions, Enlighten Not Frighten, Know Tech Not No Tech
References
1/ https://www.thewhitehatter.ca/post/deepnudes-undressing-ai-generated-intimate-image-abuse-material
