How AI is applied across API Evangelist and APIs.io. Read my AI disclosure →
API Evangelist API Evangelist
Discovery
Learnings
Guidance
Toolbox
Alignment
API Evangelist LLC

Safe Social Media Act (Bill C-34)

Bill — not yet law Canada

Introduced but not enacted. It can be amended beyond recognition or die on the order paper, and nothing in it binds anyone today. Catalogued because the obligations it proposes are the ones worth building toward.

Introduced 10 June 2026, Bill C-34 would enact the Digital Safety Act and create a Digital Safety Commission of Canada — imposing safety duties, risk assessment and transparency on large online platforms, with a takedown regime for the most serious content. Its notable feature is that it applies expressly to AI CHATBOT SERVICES alongside social media, making it the first of the online-safety regimes in this catalog to name conversational AI in its scope rather than reaching it by interpretation.

Bill C-34 is Canada’s second attempt at online-harms legislation. The first, Bill C-63, was introduced in February 2024 and drew sustained criticism for bundling a platform-safety regime together with Criminal Code and Human Rights Act amendments — including a standalone hate-speech provision that dominated the debate and eventually sank the bill. C-34, introduced 10 June 2026, drops the criminal provisions and keeps the platform regime.

What it would create:

  • The Digital Safety Act - Safety duties on regulated services: assess the risk their systems create, mitigate it, and be transparent about both.
  • A Digital Safety Commission of Canada - With enforcement powers and a mandate to support victims.
  • A takedown regime for the most serious content - Narrowly scoped, on a short clock.
  • Transparency and accountability obligations - Reporting on what the service found and what it did.
  • Scope that names AI chatbot services - Alongside social media services and certain other online services.

That scope clause is why this entry matters beyond Canada. Every other online-safety regime in this catalog reached generative AI by interpretation — Ofcom opening investigations into Grok and an AI companion service under an Act drafted before those products existed, and arguing that a model generating content into a shared surface makes the service a user-to-user service. That argument will be litigated. Canada is drafting around it by naming chatbot services in the statute, which is what a legislature does when it has watched another regulator fight for a reading and decided to spare its own the trouble.

The design lesson C-34 absorbed from C-63 is also worth recording, because it is the same lesson visible in Colorado’s AI Act and the EU’s Digital Omnibus. The duties that survive are systemic and procedural: assess, mitigate, disclose, report. The provisions that sink a bill are the ones that make a specific piece of speech unlawful. A platform-safety regime that asks about process passes; one that adjudicates content does not.

It is a bill at first reading and it binds no one. I carry it because a Canadian company reading only what is in force today would conclude that online safety is not a Canadian issue, and would then be surprised twice — once by C-34, and once by the fact that Canadian users are already covered by what platforms built for the UK and Australia. The engineering has largely been done. What changes is who can ask to see the records.

Where this applies

This is a horizontal regime: it binds companies by what they do with data, software or customers, not by the sector they sit in. Every industry below is one where it creates specific, additional duties — it is not the limit of who has to comply.

Countries
Regions