By: Emma Marck
Legislation Commented On: Protecting Victims of Non-consensual Distribution of Intimate Images Act, SA 2017, c P-26.9
PDF Version: Intimate Image Abuse in the Age of Deepfakes: Reforming Alberta’s Civil Legislation: Part 1
One in ten American teenagers personally knows someone who has had deepfake non-consensual intimate images (DNCII) of them created, and 6% reported having been victimized by DNCII (Thorn, Deepfake Nudes & Young People: Navigating a New Frontier in Technology-Facilitated Nonconsensual Sexual Abuse and Exploitation (Thorn: 2025) at 14). Recent incidents in Alberta demonstrate why intimate image legislation must address DNCII and its online distribution. In December 2025, a Calgary teen was charged in relation to DNCII of students from several schools. Six months later, two Edmonton boys were charged after creating and sharing DNCII of classmates. Reported consequences include job loss, school expulsion, post-traumatic stress disorder, and, in some cases, death by suicide (Mary Anne Franks, “‘Revenge Porn’ Reform: A View from the Front Lines” (2017) 69:5 Florida L Rev 1251 at 1259, 1263).
Section 162.1(1) of the Criminal Code, RSC 1985, c C-46 makes the non-consensual distribution of intimate images a criminal offence. Bill C-16, or the Protecting Victims Act, SC 2026, c 19, s 15(2), amended section 162.1(2)(b) to capture DNCII and came into force on July 18, 2026.
Alberta was among the first provinces to make the non-consensual distribution of intimate images (NCDII) a statutory tort under the Protecting Victims of Non-consensual Distribution of Intimate Images Act, SA 2017, c P-26.9 (Alberta Act). Since its enactment in 2017, the landscape of intimate image abuse has evolved. Perpetrators have the ability to create and distribute DNCII with increasing speed and realism. Currently, Alberta is one of the only provinces in Canada with NCDII legislation that does not capture DNCII. The Alberta Act also fails to cover “nearly nude” images of an individual toileting, dressing, undressing, or images captured through upskirting. Although nearly nude images may “appear less severe,” their non-consensual distribution can cause the same “psychological damage, reputational harm and . . . emotional distress” as NCDII already covered by the Alberta Act (Manitoba, Legislative Assembly, Debates and Proceedings, 43rd Leg, 3rd Sess, vol LXXX, No 48 (4 May 2026) at 1674 (Jodie Byram)).
In March, the Alberta government announced plans to amend the current legislation to include DNCII in its civil cause of action. Although there are many important details to examine, this post focuses on the most necessary amendments that should be included to improve access to justice for victims of NCDII and DNCII abuse. My arguments will be published as Parts 1 and 2. In this first post, I argue that the Legislature should:
1) Amend section 1 of the Alberta Act to define “intimate image” as a “personal intimate image” or a “fake intimate image” and include “nearly nude” images under each definition. Personal intimate images should require that the person is “depicted” instead of “identifiable”. Fake intimate images should maintain an identifiability requirement.
In the second post, I explore two additional arguments. The Legislature should:
2) Create a right to apply to the Court of King’s Bench for an expedited intimate image protection order. The Alberta Act should also introduce separate consent standards for expedited protection orders and damages claims.
3) Expressly authorize removal, search-engine de-indexing, information production, and direct intermediary orders. Orders involving content removal and de-indexing will be referred to collectively as “takedown” orders in this post. These reforms must be supported by an internet-intermediary definition and liability limit.
There have been no reported decisions under the Alberta Act’s civil cause of action for NCDII, and I will refer to case law from other jurisdictions to inform my analysis.
The Alberta Act Should Create a New Definition of “Fake Intimate Image”
The Alberta Act currently defines intimate images as a “visual recording of a person made by any means, including a photograph, film or video recording” (s 1(b)). Alberta’s legislature must decide whether to expand the existing definition of “intimate image” or create a separate “fake intimate image” definition including DNCII.
R v MSK, 2026 NSPC 12 (CanLII) (MSK) illustrates the gap in the Alberta Act, which currently defines an intimate image only as a “visual recording” (s 1(b)). Before the recent amendment to s 162.1(2) of the Criminal Code, an intimate image was defined as a “visual recording of a person made by any means including a photographic, film or video recording… in which the person is nude, is exposing his or her genital organs or anal region or her breasts or is engaged in explicit sexual activity”. The Court in MSK found that the AI-generated nude images at issue were not “visual recordings” under s 162.1(1) because they were “created by technology” rather than captured by a camera (at paras 60, 105–106). Judge Bronwyn Duffy called for legislative revision and refused to “force-feed” deepfake images into “incompatible legislation” (at para 105). Similarly, the Court in R v R.K.1, 2025 ONCJ 542 (CanLII) (R.K.1) found that the words “the person” and “exposing his or her” indicated a legislative intent to protect only authentic recordings of an individual’s actual body parts (at para 27).
“Altered” Images May Not Capture Wholly Generated AI Images
DNCII includes both modifications of identifiable source material and the creation of entirely new depictions. At the time of the Uniform Law Conference of Canada’s report on the Nonconsensual Disclosure of Intimate Images (NCDII) Tort (ULCC Report), the discussion focused on manipulating existing images such as face-swapped deepfakes (at paras 34–38). The ULCC Report recommended excluding “wholly original content” from the definition of intimate images in part because it viewed the resulting harms as different and identified artistic and freedom-of-expression concerns (at para 38).
British Columbia’s Intimate Images Protection Act, SBC 2023, c 11 (BC IIPA) and Saskatchewan’s The Privacy Act, RSS 1978, c P-24 (SK Privacy Act) provide that an intimate image may qualify for statutory remedies whether or not the image “has been altered in any way” (BC IIPA, s 1; SK Privacy Act, s 7.1). The Government of British Columbia’s guidance on intimate images clarifies that the BC IIPA covers fake, altered, and AI-generated intimate images, including deepfakes (BC IIPA, s 1). The phrase “altered in any way” captures manipulations of an existing image, including face swaps and nudification. However, it is not clear whether this wording captures a wholly generated image where the original image itself is not identifiable or does not exist (Suzie Dunn, “Legal Definitions of Intimate Images in the Age of Sexual Deepfakes and Generative AI” (2024) 69:4 McGill LJ 395 at 412–413 (Dunn)).
Manitoba reduces this uncertainty by defining a “fake intimate image” to include images created through “software, machine learning, artificial intelligence or other means, including by modifying, manipulating or altering an authentic visual representation” (The Non-Consensual Distribution of Intimate Images Act, CCSM c N93, s 1(1) (Manitoba Act)). This provides a harm-centred approach by recognizing that fake intimate images can be “used to humiliate, extort or manipulate victims without any original image ever being shared” (Manitoba, Legislative Assembly, Debates and Proceedings, 43rd Leg, 3rd Sess, vol LXXX, No 48 (4 May 2026) at 1675 (Jodie Byram)). The Alberta Act should adopt Manitoba’s separate fake intimate image category to capture both authentic recordings and DNCII in its definitions.
The Manitoba Act does not define “artificial intelligence.” Although Bill 51, The Public Sector Artificial Intelligence and Cybersecurity Governance Act, SM 2026, c 43, defines an “artificial intelligence system,” its definition does not apply to Manitoba’s intimate-image legislation. The contextual placement of “artificial intelligence” alongside “software,” “machine learning,” and the phrase “other means” may reduce the need for a fixed definition and allow the fake intimate image definition to respond to evolving methods of creating synthetic intimate images (Manitoba Act, s 1(1)).
Identifiability Should Not Apply to Personal Intimate Images
The ULCC warned against an identifiability requirement as part of an intimate image definition. Identifiability focuses on reputational harm rather than invasions of privacy, loss of sexual autonomy, sexual objectification, emotional distress, and fear of future identification (ULCC Report at para 52). Identifiability asks whether third parties can recognize the individual from the image or information connected to it. However, requiring third-party identifiability for personal intimate images requires public recognition to obtain relief and could force an individual to wait until “the worst damage possible” has occurred (ULCC Report at paras 44, 51–52). In contrast, depiction requires proof that the image is of or represents the individual, regardless of whether third parties can personally recognize the individual (ULCC Report at paras 44, 48, 50–51). Depiction may extend to an image showing “any part of the body” (ULCC Report at para 50).
British Columbia’s definition of “intimate image” requires only that the individual be depicted. The BC IIPA applies where an individual “is or is depicted” in a “visual recording or visual simultaneous representation of an individual, whether or not the individual is identifiable” (BC IIPA, s 1). The BC IIPA separately defines “depict” as meaning “to represent in” a “recording” or “simultaneous representation” (BC IIPA, s 1).
The distinction between identifiability and depiction was relevant in two cases from British Columbia. In AQ v BW, 2025 BCCRT 907 (CanLII), which concerned a close-up photograph of AQ’s groin and upper thigh (at para 19). In an unpublished decision regarding AQ’s application for protection orders, British Columbia’s Civil Resolution Tribunal (CRT) found that the image depicted AQ (at paras 2, 19). The CRT subsequently awarded damages despite finding that AQ was “not identifiable from the image itself.” (at paras 28, 30)
In contrast, in MN v OP, 2025 BCCRT 750 (CanLII), the CRT applied an objective test for depiction, asking whether a “reasonable observer” could “perceive the person in the representation as [the applicant]” (at para 22). It found that the image did not depict the applicant. The CRT could not locate the birthmark in the image that initially led MN to believe the depiction was of herself, and the image had not been altered to resemble the applicant (at para 21).
I agree with the ULCC’s recommendation and suggest that Alberta’s legislation should use the element of depiction instead of identifiability when defining a “personal intimate image”. A personal intimate image has an independent factual connection to the person because the image records an individual’s own body or sexual activity (ULCC Report at paras 44, 48). For example, a person may know that an intimate image shows their body despite nobody else recognizing or attributing the photograph to them (ULCC Report at para 44). An objective test for depiction should not be applied.
Identifiability Should Apply to Fake Intimate Images
Identifiability performs different functions for personal and fake intimate images. The ULCC Report explained why identifiability should not be required for personal intimate images (ULCC Report at paras 44, 48, 50–52). The Criminal Code and the Manitoba Act demonstrate that identifiability should be required for fake intimate images because it performs a different limiting function for DNCII (Criminal Code, s 162.1(2)(b); Manitoba Act, s 1(1)).
Deepfakes cause distinct harm because DNCII can falsely sexualize a person without recording an intimate moment that actually occurred (Dunn at 413). The ULCC Report recognized that DNCII can cause emotional harm, reputational harm, loss of employment, stalking, harassment, sexual objectification, shame, humiliation, and loss of sexual agency (ULCC Report at paras 34–38). Identifiability connects the false depiction to a real person and limits generic, fictional, or wholly unconnected sexual content from the fake intimate image definition (Dunn at 414). For example, Manitoba distinguishes “personal intimate images” that do not contain an identifiability requirement from “fake intimate images” that falsely depict an identifiable person (Manitoba Act, s 1(1)). Dunn treats identifiability and depiction as cumulative requirements, explaining that minor technical flaws should not exclude an image “so long as the person is identifiable and the image has a clear realism to it.” (Dunn at 414)
Bill C-16’s amendment to the Criminal Code definition of “intimate image” supports a bifurcated approach. Section 162.1(2) of the Criminal Code now defines intimate images as either a “visual recording” or a “visual representation.” A “visual recording” under s 162.1(2)(a)(iii) requires that “the person depicted retains a reasonable expectation of privacy at the time the offence is committed.” A “visual representation” under s 162.1(2)(b) must be “made by any electronic or mechanical means, including by means of artificial intelligence software,” and show “an identifiable person who is depicted as nude, as nearly nude, as exposing their sexual organs or as engaged in explicit sexual activity.” Section 162.1(2)(b) also requires that “the depiction [of an identifiable person] is likely to be mistaken for a visual recording of that person.”
The Department of Justice Canada’s Charter Statement for Bill C-16 recognized that the amendment limits the communication of expressive content and engages section 2(b) of the Canadian Charter of Rights and Freedoms. However, the meaning of “visual representation” is carefully tailored to exclude images that depict unidentifiable or fictitious persons or are unlikely to be mistaken for authentic recordings.
The National Association of Women and the Law (NAWL), argued in its brief on Bill C-16 that harm from DNCII arises from the exploitation of a person’s image to produce sexual material, and not whether others believe the material is an authentic recording (at 6). Accordingly, NAWL argues that requiring depiction to be “likely to be mistaken for a visual recording of that person” is too restrictive as a definition (at 6). For example, DNCII depicting a person engaging in sexual activity with an imaginary creature or in outer space may cause harm even if the depiction does not satisfy a test of realism. NAWL reasoned that it is sufficient to require that DNCII represents an identifiable person (at 6; see also Dunn at 414).
For the purposes of Bill C-16, NAWL recommended requiring the person’s body to be depicted in a “reasonably convincing manner” to exclude cartoonish representations (at 6). In this sense, NAWL recommends that “relative realism” should focus on the body of the person instead of the context of the image (at 6). This approach limits the definition to representations that appropriate a real person’s sexual identity while including depictions placed in unrealistic contexts (at 6–7). The Manitoba Act adopts this approach, requiring that a fake intimate image must “falsely depic[t] an identifiable person” “in a reasonably convincing manner” (s 1(1)).
Alberta’s legislation can best address wholly generated images by bifurcating its existing definition to include a “fake intimate image”, defined as a visual representation created through “software, machine learning, artificial intelligence or other means, including by modifying, manipulating or altering an authentic visual representation” that, “in a reasonably convincing manner, falsely depicts an identifiable person”, like in the Manitoba Act (s 1(1)). Classifying personal intimate images separately from fake intimate images, without requiring identifiability, enables a cause of action for both reputational harms and invasions of privacy (ULCC Report at para 52). Alberta’s legislation can define a personal intimate image as a “visual recording of a person, whether or not the person is identifiable, who is engaging in a sexual act, is nude or nearly nude, or exposes their genital organs, anal region, or breasts …”. Canadian intimate image statutes have historically referred to the “person depicted in the image” without expressly requiring that the person be identifiable (ULCC Report at para 48).
The Alberta Act Should Require That It Be “Reasonable to Suspect Non-Consent” for Fake Intimate Images
The issue in MSK involved whether a reasonable expectation of privacy (REP) could extend to fake intimate images. However, the Court’s reasoning was based on the non-consensual creation and distribution of false sexual depictions using the complainants’ likenesses (MSK at paras 64–69). The Court in MSK found that manipulating publicly available photographs “without their knowledge or consent” deprived the complainants of control over their images and breached their sexual integrity (MSK at para 67).
Requiring REP for fake intimate images invites victim-blaming by directing legal scrutiny toward a person’s “‘risky’ or ‘immodest’ behaviour” as diminishing their privacy expectations (Moira Aikenhead, “A “Reasonable” Expectation of Sexual Privacy in the Digital Age” (2018) 41:2 Dal LJ 273 at 289–90). A consent-based definition instead focuses on the non-consensual and sexualized use of the depicted person’s likeness. This approach better fits DNCII because the privacy violation does not depend on whether the complainant had a REP in an authentic recording.
Manitoba distinguishes fake intimate images from personal intimate images in this context. For “personal intimate images”, the depicted person must have a REP when the image is created and, if the image is distributed, when it is distributed. The Alberta Act should follow Manitoba’s bifurcated approach by requiring a REP for personal but not fake intimate images. For fake intimate images, it must be “reasonable to suspect that the person depicted in the image would not consent to the recording being made or distributed to others” (Manitoba Act, s 1(1)(c)).
Section 5 of the Alberta Act confirms that a depicted person does not lose REP merely because the person consented to the recording or provided the image to another person where the recipient knew or ought reasonably to have known that the image was not to be distributed further. The existing REP element should be retained within the proposed “personal intimate image” definition to reflect images recorded in circumstances that actually occurred (Alberta Act, s 1(b)(ii)–(iii); ULCC Report at paras 44, 48).
The Alberta Act Should Include “Nude or Nearly Nude” Images Under Both Definitions of “Intimate Image”
Nearly nude images of an individual toileting, dressing, undressing, or captured through upskirting cause harms similar to those caused by nude intimate images and involve the same blameworthy conduct (ULCC Report at para 43). Section 1(b)(i)(A) of the Alberta Act currently defines an “intimate image” to depict a person who is “nude,” “exposing genital organs … anal regions or breasts,” or engaging in “explicit sexual activity.”
Roque v Peters, 2022 MBQB 34 (CanLII) (Roque) was decided before Manitoba amended its definitions to qualify images capturing a person who is “nude or nearly nude” as a fake intimate image or a personal intimate image in 2026 (Manitoba Act, s 1(1)(a)(ii)). The Court found that 16 of the 21 images at issue in that case did not meet the definition of intimate image. This finding was made despite the fact that the images showed the plaintiff “in various stages of undress or in her underwear” and that they were taken in circumstances giving rise to a reasonable expectation of privacy (at para 40). Manitoba subsequently amended the definition to include “nearly nude” images in 2026 (The Non-Consensual Distribution of Intimate Images Amendment Act, SM 2026, c 2, s 2(1) (Manitoba Amendment Act)).
The CRT interprets “nearly nude” in B.D.S. v M.W., 2024 BCCRT 410 (CanLII) (B.D.S.). The amount of clothing worn and the context in which the image was created inform whether a person is “nearly nude” (at para 36). Although the applicant was wearing underwear that was “not more revealing than many swimsuits,” the image was found to be “nearly nude” because anyone viewing the photo would immediately recognize that it depicted “a sexualized and private moment.” (at paras 37, 39) This interpretation demonstrates how including “nearly nude” images could capture some images excluded in Roque under the Manitoba Act’s former definition.
Alberta should address the gap in its legislation by including “nude or nearly nude” in the proposed definitions of “intimate image”, both fake and personal. In addition, REP requirements should limit the scope of “nearly nude” for personal intimate images (ULCC Report at para 42). Requiring that it be reasonable to suspect non-consent would similarly limit the scope of “nearly nude” for fake intimate images (see e.g. Manitoba Act, ss 1(1)(a)(ii)–(c); Manitoba Amendment Act, s 2(1)(a)).
Conclusion
In summary, Alberta’s legislation should expand the definition of “intimate image” to include personal intimate images and fake intimate images. A personal intimate image definition would protect privacy and sexual autonomy even where the person is not identifiable to third parties. A fake intimate image definition would capture wholly generated images while using identifiability as a necessary limit on expressive content. Including “nude or nearly nude” images in both definitions would prevent affected persons from being denied relief because the image falls just outside Alberta’s current wording. Accordingly, the provisions could be drafted as:
“fake intimate image” means a visual recording that is created through the use of software, machine learning, artificial intelligence or other technological means, including by modifying, manipulating or altering an authentic visual representation
(i) that, in a reasonably convincing manner, falsely depicts an identifiable person who
(A) is nude, nearly nude or is exposing his or her genital organs or anal region or her breasts, or
(B) is engaged in explicit sexual activity,
(ii) in respect of which it is reasonable to suspect that the person depicted in the image would not consent to the recording being made or distributed to others.
“intimate image” means
(a) a personal intimate image of a person; or
(b) a fake intimate image of a person.
“personal intimate image” means a visual recording of a person made by any means, including a photograph, film or video recording, whether or not the person is identifiable
(i) in which the person depicted in the image
(A) is nude, nearly nude, or is exposing his or her genital organs or anal region or her breasts, or
(B) is engaged in explicit sexual activity,
(ii) which was recorded in circumstances that gave rise to a reasonable expectation of privacy in respect of that image, and
(iii) if the image has been distributed, in which the person depicted in the image retained a reasonable expectation of privacy at the time it was distributed.
This post may be cited as: Emma Marck, “Intimate Image Abuse in the Age of Deepfakes: Reforming Alberta’s Civil Legislation: Part 1” (25 August 2026), online: ABlawg, http://ablawg.ca/wp-content/uploads/2026/08/ Blog_EM_Intimate_Images_Part1/.pdf
To subscribe to ABlawg by email or RSS feed, please go to http://ablawg.ca
Follow us on Twitter @ABlawg