A growing tension between national security and civil liberties is coming to a head in Canada, as proposed legislation threatens to expand state access to private digital communications including interactions with artificial intelligence platforms.
At the heart of the debate is a fundamental question: how much privacy must citizens sacrifice in the name of safety?
Advocates for stronger surveillance powers argue the answer is straightforward. Governments with broader access to phone calls, emails, texts, and now AI conversations could theoretically catch more criminals, prevent more violence, and save more lives. But critics push back hard, insisting that this logic misses something essential about what it means to live in a free society.
“Arguing that you don’t care about privacy because you have nothing to hide is like arguing that you don’t care about free speech because you have nothing to say,” whistleblower Edward Snowden famously observed a quote that privacy advocates continue to invoke when challenging the creeping expansion of state surveillance.
Canada’s Section 8 of the Charter of Rights and Freedoms explicitly protects citizens against unreasonable search and seizure. For decades, this provision has required law enforcement to obtain a judicial warrant before monitoring a person’s communications a safeguard that defenders say strikes a careful balance between fighting crime and preserving freedom.
Under the existing warrant system, police can access the emails, texts, and internet activity of suspected criminals, but only with court authorization. That system, supporters argue, has served Canada well for generations allowing investigators to pursue genuine threats while leaving ordinary citizens alone.
The concern now is that new legislation may quietly dismantle that balance.
The proposed Lawful Access Act, known as Bill C-22, has drawn fierce criticism from legal scholars and civil liberties groups. If passed, it could classify AI companies like OpenAI as “electronic service providers,” opening the door for law enforcement to access user data under a lower legal threshold than currently required.
Crucially, the bill would shift the standard from “reasonable grounds to believe” to the less demanding “reasonable grounds to suspect” a change critics say makes it significantly easier for police to obtain sensitive subscriber information without robust judicial oversight.
The legislation would also compel electronic service providers to retain user metadata including IP addresses, connection times, and browsing frequency for up to one year, making it available to authorities subject to a warrant. University of Ottawa law professor Michael Geist has described mandatory metadata retention as one of the most privacy-invasive tools a government can deploy, warning that it creates surveillance capabilities that extend well beyond any stated anti-crime objective.
The renewed push for AI surveillance gained momentum following a deadly shooting in Tumbler Ridge, British Columbia, in February. Eight months before the attack, the shooter, Jesse Van Rootselaar, had used OpenAI’s ChatGPT to explore scenarios involving gun violence. OpenAI detected the activity and banned him from its platform, but did not report the matter to police, determining there was no credible or imminent threat.
What makes the case more complicated is that police had previously removed firearms from Van Rootselaar’s home in 2024 and then returned them less than a month before the shooting, also seeing no immediate danger.
Critics of expanded AI surveillance argue that the Tumbler Ridge tragedy illustrates something important: the existing warning signs were visible to both OpenAI and law enforcement, yet neither acted on them in time. Placing AI companies under state surveillance, they contend, would not necessarily have changed that outcome and would come at a steep cost to the freedoms of millions of law-abiding Canadians.
Historians and legal scholars frequently point to the 20th century as a cautionary tale about what happens when governments accumulate unchecked surveillance power. Authoritarian regimes from Stalin’s Soviet Union to Nazi Germany and Mao’s China used state surveillance not to protect their citizens, but to control, silence, and ultimately destroy them. The death tolls running into the tens of millions surpassed even the catastrophic military casualties of two world wars.
“Repressive regimes spy on citizens to ensure obedience and compliance,” wrote John Carpay, President of the Justice Centre for Constitutional Freedoms. “Citizens who are spied on by the authorities are not truly free, even if they might superficially enjoy some freedom of speech.”
Beyond the legal arguments, some observers insist that mass violence in Canada cannot be solved through surveillance alone. They point instead to a cluster of social and cultural problems family breakdown, social isolation, declining mental health, and eroding community bonds as the deeper drivers of violence.
Monitoring what Canadians type into AI chatbots, they argue, does nothing to address any of those root causes. Meanwhile, the chilling effect on free expression could be significant: if citizens know their AI queries may be seen by the government, they may self-censor, avoid exploring sensitive ideas, or disengage from the kind of open intellectual inquiry that democracy depends on.
The outcome of the Bill C-22 debate will likely shape the boundaries of digital privacy in Canada for years to come. At its core, the question is not simply about AI or metadata or warrant thresholds. It is about what kind of society Canada wants to be one where the state is trusted with sweeping access to private life, or one where freedom, including the freedom to think privately, remains genuinely protected.
As the legislative process unfolds, that question grows more urgent by the day.
