What is Bill C-34?

Bill C-34 is the Liberal government’s latest attempt to regulate speech online, or as they put it, to prevent “harmful” speech.

The bill would create a Digital Safety Commission of Canada, which would consist of three to five commissioners whose job would be to police speech for all Canadians, supported by a new bureaucracy.

The Digital Safety Commission would enforce:

• a “duty to act responsibly” for social media companies and artificial intelligence chatbot companies by mitigating the risk that Canadians are exposed to “harmful content” 
• a “duty to protect children” for social media companies to keep children under age 16 off social media services and pornographic websites; and
• a duty for social media companies to remove content that “sexually victimizes a child or revictimizes a survivor and intimate content communicated without consent.”

Internet companies that do not comply would face fines of up to $10 million or 3% of global revenue.

While the CCF supports keeping kids safe, this bill is a sneak-attack on the constitutionally-protected right to freedom of expression for adults and mature minors. 

Bill C-34 shifts decisions about acceptable public discourse away from citizens and toward a handful of bureaucrats in Ottawa.

Most shockingly, the bill would censor the output of artificial intelligence chatbots like ChatGPT and Gemini to ensure they don’t give “harmful” answers on sensitive topics. 

This may be the greatest threat to free speech in decades. 

What “harmful content” would social media and AI companies be required to “mitigate”?

Bill C-34 defines harmful content as:

• (a) intimate content communicated without consent;
• (b) content that sexually victimizes a child or revictimizes a survivor;
• (c) content that induces a child to harm themselves;
• (d) content used to bully a child;
• (e) content that foments hatred;
• (f) content that incites violence; and
• (g) terrorism or violent extremism content.‍

What counts as “content that foments hatred” and why would a duty to mitigate this risk freedom of expression?

Content that foments hatred is defined in the Act as “content that expresses detestation or vilification of an individual or group of individuals on the basis of a prohibited ground of discrimination, within the meaning of the Canadian Human Rights Act, and that, given the context in which it is communicated, is likely to foment detestation or vilification of an individual or group of individuals on the basis of such a prohibited ground.‍”

Requiring companies to “mitigate” exposure to this is an enormous risk to freedom of expression. Here’s why.

First, “hatred” has been defined by the courts as only the “extreme” manifestations of detestation and vilification. The bill, as currently written, does not require the content to be “extreme.” The government already tried to water down the definition of “hatred” in Bill C-9, and they were told by the CCF and many others that if they want to define hatred, they must use the definition that tracks the language used in case law. The government in C-9 did reverse course, but now they’re trying again to water down the definition knowing full well what they are doing may violate the Constitution.

Second, “vilification and detestation” are extremely subjective concepts. Judges and lawyers have difficulty deciding what is over the line, and there is very little guidance from courts on what these terms mean. Now a bureaucrat and tech company employees will be trying to apply these loose terms.

In the Supreme Court decision Saskatchewan (Human Rights Commission) v. Whatcott, Justice Rothstein defined detestation as “tend[ing] to inspire enmity and extreme ill-will against” a group which “goes beyond mere disdain or dislike,” and vilification as “seek[ing] to abuse, denigrate or delegitimize … to render lawless, dangerous, unworthy or unacceptable.” What does that mean?

Unclear. The Court has also said that we can recognize hatred by looking for “hallmarks of hatred,” including speech that:

• “vilifies the targeted group by blaming its members for the current problems in society”
• “alleg(es) that members of a group are a ‘powerful menace’”
• “accuses a group of carrying out secret conspiracies to gain global control or that they are plotting to destroy western civilization”
• “suggests members are illegal or unlawful,” such as by labelling them “liars, cheats, criminals and thugs” 
• calls people “pure evil” 
• “equates the targeted group with groups traditionally reviled in society, such as child abusers, pedophiles or deviant criminals who prey on children”
• describes members of a group as “animals or as subhuman” 
• “calls into question whether group members qualify as human beings” or 
• refers to them as “horrible creatures who ought not to be allowed to live,” “incognizant primates,” “genetically inferior,” “lesser beasts” or “sub-human filth.”

If judges and lawyers have difficulty determining whether a particular statement counts as hatred, social media companies and AI companies concerned about facing fines will likely err on the side of censoring more expression by blocking it or filtering it out.

The result would be a sanitized social media experience where people cannot have difficult conversations, and sanitized responses from AI chatbots that would shy away from difficult truths or unpopular perspectives on topics ranging from immigration to gender identity.

The Supreme Court has been clear that freedom of expression exists to protect all expression, and in particular expression that is distasteful, unpopular, or contrary to the mainstream. Otherwise, people cannot get to the truth of matters and we cannot govern ourselves as a democracy.

What counts as a social media service?

This is not entirely clear. A social media service is defined in the Act as one with “a number of users that is equal to or greater than the significant number of users provided for by regulations” or that is “designated as a regulated social media service by regulations.” In other words, this will be decided later through regulation. The Cabinet will decide which social media companies the bill applies to, at their discretion, and without debate.

What is clear is that private messaging features are excluded. “Private messaging feature” is defined as one that “enables a user to communicate content to a limited number of users determined by the user; and does not enable a user to communicate content to a potentially unlimited number of users not determined by the user.”

How does the government plan to keep children under 16 off of social media? 

The Bill would create a “duty to protect children” including by requiring social media companies to “implement age-verification or age-estimation measures designed to prevent a person under the age of 16 from being able to have an account with, or be otherwise registered with, the service.”

The Digital Safety Commission will decide whether the measures are “effective” and the social media company would be required to “provide for the destruction of personal information that is collected for age-verification or age-estimation purposes once the verification or estimation is completed.”

Age-verification measures may include things like requiring all adults to provide a driver’s license or other form of ID to create an account. This raises privacy concerns.

Age-estimation measures may involve artificial intelligence guessing a person’s age based on their appearance. In other countries, kids have gotten around these measures by using fake moustaches.

Does keeping kids off social media raise concerns about freedom of expression?

Yes. The Charter guarantees “everyone” the right to freedom of expression.

There is no question this includes people under the age of 16. 

People under 16 who wish to express themselves on social media by sharing their art, their writing, their opinions or promoting a business will no longer be able to do so. 

In law, there are many things that mature minors who are 14 or 15 have the right to do without government interference. It is unknown at this time whether this limit on expression would be upheld by the courts. 

While there are age verifications for all kinds of things (buying pornographic magazines, cigarettes, entering bars), the analogy doesn’t work when applied to the Internet. Stores don’t keep records of everything you buy, say or do inside the store for years. You aren’t required to show ID that is stored at every building you enter. But that’s what age verification online does. Right now we don’t know what the age verification will look like. The verification plans will be developed by the platforms and approved by the new Digital Safety Commission, but Canadians deserve privacy online and whatever is proposed needs to comply with the Charter and privacy legislation.

The ability to use the Internet anonymously has been vital to using it for political dissent. People use the internet to engage in journalism and advocacy, and all kinds of things the government may not like. We don’t know what age verification will look like yet, but we must push back on the idea that keeping kids safe requires adults to show their driver’s license so they can scroll X. 

Would this bill require social media companies to take down content?

Yes. Social media companies would be required to take down two types of content proactively: (1) content that the operator has reasonable grounds to suspect is content that sexually victimizes a child, and (2) content that the operator has reasonable grounds to believe “revictimizes a survivor or intimate content communicated without consent.”

This is a good idea. 

And it is limited to those specific types of content. 

It does not appear that the Digital Safety Commission would make orders requiring social media companies to take down other specific instances of “harmful content.” 

However, companies that do not “mitigate exposure to” other forms of harmful content would face enormous potential fines. That’s why we are concerned the bill is a sneak-attack on free expression. The government knows it cannot directly ban unpopular, rude, divisive, transgressive or even offensive speech. So it is outsourcing this censorship role to online platforms by creating a regulatory chill. But a government that cannot constitutionally ban speech outright should not achieve the same result indirectly by pressuring private platforms through a regulator. Platforms will likely take down this “lawful but awful” (or even just unpopular) content on their own accord to avoid potential liability.

Would this bill require AI companies to not give “harmful” answers?

Yes. AI companies would be required to “implement measures that are adequate to mitigate the risk that the service will communicate harmful content to a user of the service.” This includes “hate” content. There are all kinds of questions that may not in fact “foment hatred”, but that an AI company may decide are too risky to answer because the answer could be interpreted as encouraging a negative view of a protected group. People can disagree about whether a particular statement:

• criticizes an ideology,
• criticizes a religion,
• criticizes immigration policy,
• criticizes government Indigenous policy,
• or actually crosses the line into hatred.

Once a regulator is involved, there will be pressure on platforms and AI companies to take a cautious approach. This will almost certainly lead to chatbots refusing to engage on controversial topics out of fear of regulatory reprisal.  

The bill would also require AI companies to mitigate the risk of harmful behaviour including:

• posing as a human being in a manner likely to lead a user of the service to mistake it for a human being or otherwise being deceptive about being an artificial intelligence system;
• posing as a medical, legal or other licensed professional and giving advice based on that deception that could reasonably be expected to be relied on by a user of the service;
• using manipulative engagement techniques to encourage a user of the service to form or maintain an emotional attachment to the service in a way that may encourage the user to withdraw socially or disconnect from reality;
• encouraging self-harm, suicide or the commission of acts that could cause death or serious bodily harm to an individual; or
• any other type of behaviour specified in the regulations.

This would fundamentally change the AI experience, as users would no longer know whether AI is giving them truthful answers or answers filtered to be politically correct and not offside the law.

What can I do?

Sign and share our petition: theccf.ca/stopc34/