Back to Bills

Regulating the Online Use of Deepfakes Act

Full Title:
An Act to provide for the regulation of the online use of deepfakes and for related transparency measures

Summary#

Bill C-277 would set rules for how online platforms handle “deepfakes” (AI-edited or AI-generated images, audio, or video that look or sound real and use someone’s appearance or voice). It requires platforms to detect, label, and remove deepfakes that could cause significant harm. It also creates an administrative (non-criminal) penalty system to enforce these duties. The stated goal is to protect people from emotional, reputational, and financial harm and to increase transparency online.

Key changes:

  • Platforms and online services accessible in Canada must set up ways to identify deepfakes, including user reporting tools.
  • They must take reasonable steps to clearly label deepfakes shared on their services.
  • They must take reasonable steps to prevent deepfakes that may cause “significant harm” from being shared, and remove such content as soon as feasible after it is identified.
  • They must keep records needed to show compliance. The federal cabinet can set detailed record-keeping rules.
  • An administrative monetary penalty system is created. The cabinet will set which violations can be penalized, penalty amounts or ranges, procedures, possible defences, and appeal/review processes.
  • Company directors and officers who direct or take part in a violation can be personally liable for penalties.
  • Violations are not criminal offences.

What it means for you#

  • Users

    • You should see clearer labels on content identified as deepfakes.
    • You will have reporting tools to flag suspected deepfakes.
    • Harmful deepfakes about you or others should be removed once identified.

  • People targeted by deepfakes

    • Platforms must act to prevent sharing of deepfakes that may cause significant harm (for example, identity theft, job loss, humiliation).
    • Harmful deepfakes must be taken down as soon as feasible after they are identified.

  • Creators and posters of content

    • If you share AI-altered or AI-generated media using a real person’s image or voice, platforms will likely require clear labeling.
    • Content that may cause significant harm could be blocked or removed.

  • Online platforms and services (including social media)

    • You must build and operate systems to detect deepfakes, allow user reports, label deepfakes, and prevent and remove those that may cause significant harm.
    • You must keep records to show compliance. Details may be set by future regulations.
    • You face administrative penalties for violations. Directors and officers can be personally liable if they took part in the violation.
    • The Act uses “reasonable steps,” but does not define exactly what is required. What counts as “reasonable” is unclear and may depend on future regulations.

  • Government

    • The Minister of Industry would oversee the scheme and can designate people to run it.
    • The federal cabinet will set many key details by regulation (penalty amounts, procedures, appeals, defences, and who enforces).

Expenses#

No publicly available information.

Possible cost areas:

  • Platforms may face costs to build or upgrade detection tools, reporting systems, labeling features, moderation processes, and record-keeping.
  • Platforms may face penalties for non-compliance; amounts are not yet specified.
  • The government may face costs to develop regulations and enforce the scheme; no estimates are provided.

Proponents' View#

  • The bill appears intended to reduce real harms from deepfakes, including fraud, identity theft, harassment, and reputational damage.
  • It could increase transparency by requiring clear labeling of deepfakes so people know when content is synthetic.
  • It places responsibility on platforms—the places where deepfakes spread—to detect, label, and quickly remove harmful content.
  • Using administrative penalties (not criminal charges) could promote faster, more flexible compliance.
  • The preamble states current laws do not adequately protect people from deepfake harms; this bill could fill those gaps.

Opponents' View#

  • The bill does not clearly define what “reasonable steps” look like. This may create uncertainty for platforms and uneven enforcement.
  • “Significant harm” includes psychological and reputational harm, which can be subjective. Platforms may remove borderline content to avoid penalties, which could affect satire, art, or political speech.
  • The duty to “prevent” harmful deepfakes may be read as requiring proactive scanning or monitoring. This could raise privacy concerns or lead to over-removal if detection tools are imperfect.
  • Many core elements (penalty amounts, defences, procedures, appeals, enforcement roles) are left to future regulations. Until these are set, it is hard to judge the full impact and fairness.
  • Detection technology can make mistakes. Mislabeling or removing authentic content is a possible risk.
  • Compliance could be costly or difficult for smaller services that lack in-house AI and moderation teams. The bill does not set size-based thresholds or exemptions.