Scoring Canada's Safe Social Media Act: Bill C-34 Protects; It Does Not Give

Fiscal and implementation implications

The bill creates a regulator that no current public-finance document funds. The Parliamentary Budget Officer costed C-63's Digital Safety Commission, Ombudsperson, and Office at $201 million over 2024-25 to 2028-29 with 330 full-time staff at capacity, reaching a run rate near $57 million a year; Budget 2024 provisioned $52 million over five years for the whole online-harms framework, about a quarter of that estimate. Budget 2025 contains no online-harms measure, the 2026-27 Main Estimates contain no Digital Safety Commission line, and the PBO's review of Supplementary Estimates (A) records the only current Canadian Heritage item as $6 million for the FIFA World Cup. C-34's Commission has one body where C-63 had three, and more work: AI chatbot oversight, the under-16 restriction, and an exemption-assessment function C-63 never carried. The Commission can levy cost-recovery charges on operators (s. 125), which may shift the burden from taxpayers to platforms; no charge structure exists yet.

Implementation is sequenced behind regulations the bill does not contain. The bill sets no numeric user thresholds, no commencement dates, and no compliance deadlines; which services are regulated, which platforms face the under-16 restriction, and what design features are required all await Governor in Council regulations. Minister Miller put the Commission at 18 months from royal assent (Global News, Radio-Canada); officials put the designation step at six to eight months, and the Canadian Press reported the consequence: the age restriction could bite before the exemption mechanism that is supposed to temper it exists. Michael Geist's same-day analysis reads the structure as a "trust us" bet: the operative content of the regime sits in instruments cabinet has not yet written.

The compliance bill lands on platforms and, through them, on users. The closest costed analogue, the UK's Online Safety Bill impact assessment, put ten-year present-value age-assurance costs at £17.9 million to £89.6 million on a 10-pence-per-check assumption, inside a £2.5 billion total dominated by content moderation. No Canadian regulatory impact statement exists because no regulations exist. The implication: until a funding instrument and the regulations appear, the bill's duties are commitments on paper, and the three-attempt history of this file (a 2021 consultation, C-63 dead on prorogation, now C-34) is the base rate a reader should price.

Sovereignty and trade implications

The bill asserts Canadian jurisdiction over foreign platforms' product design, and Canada's medical establishment framed it exactly that way: the CMA's same-day statement says it is "unacceptable for foreign-owned platforms to continue to get rich at the expense of our children's mental health, privacy and personal safety." The regulated population is almost entirely foreign-headquartered; Miller named Facebook, X, and Snapchat as covered (BetaKit), and officials said designation thresholds will be modelled on the Online Streaming Act and Online News Act.

Canada is converging on its peers while citing none of them. The under-16 restriction follows Australia's Social Media Minimum Age Act (in force December 10, 2025), the duty-of-care core parallels the UK's Protection of Children Codes (in force July 25, 2025), the EU published DSA minors-protection guidelines in July 2025, and the US Senate passed COPPA 2.0 by unanimous consent in March 2026; a full-text check of the C-34 release, backgrounder, and bill page finds no reference to any of them. Politico's reporting ties the tabling to the June 15 G7 summit, where the Prime Minister is pressing allies on online child safety against US and UK resistance to France's proposed common framework.

The trade exposure runs through age verification. Geist's companion analysis notes the US "strongly opposes regulations that require or create conditions that compel platforms to collect government-issued IDs." The bill's own text pushes the other way on data sovereignty: age-assurance measures qualify as adequate only if they collect nothing beyond what age-checking requires and destroy the data once the check completes (ss. 22(2), 27(2)), and the Commission cannot issue age-verification guidelines without consulting the Privacy Commissioner (s. 122(2)). The implication: whether C-34 becomes a sovereignty instrument or a trade liability depends on whether the destruction-and-minimization conditions survive into the regulations that will actually govern implementations.

Labour and economic implications

The direct employment object is a 330-person-scale regulator, by the only available proxy (PBO, 2024, staffing benchmarked on Ofcom and Australia's eSafety Commissioner). Beside it sits a compliance industry the bill summons into the Canadian market: Australia's Age Assurance Technology Trial evaluated more than 60 solutions from 48 vendors, and every regulated service in Canada will be buying from that market or building in-house.

The revenue at stake explains the platform reaction. The only independent peer-reviewed estimate, a Harvard T.H. Chan team model in PLOS ONE, put six platforms' advertising revenue from US users under 18 at nearly $11 billion in 2022, with minors accounting for 41.4% of Snapchat's ad revenue, 35% of TikTok's, and 27% of YouTube's; no Canadian equivalent exists, which this note records as a finding. Meta's published position opposes under-16 bans and asks that age verification sit with app stores (Meta Australia policy blog); Snap's CEO published an op-ed calling Australia's restriction "a massive experiment with high stakes" after Snap actioned more than 415,000 Australian accounts (Snap newsroom).

The institutional labour bearing today's harm load is also on the record. Fourteen Ontario school boards and schools are suing Meta, Snap, and TikTok for more than $8 billion over compulsive-design harms in classrooms, and the Ontario Superior Court declined to strike the claims: "It is arguable that an addictive product that interferes with the mental health and educational aspirations of students is a public nuisance that requires a remedy" (2025 ONSC 1499). The implication: the bill reallocates a burden that courts, school boards, and tip lines currently carry; whether it reduces that burden or merely renames it depends on the enforcement record the Commission builds.

Community and connection implications

The under-16 restriction withdraws a connection channel from roughly the 13-to-15 cohort, and the early Australian record measures both sides of that withdrawal. The eSafety Commissioner's March 2026 compliance update reports parent-surveyed account ownership among 8-to-15-year-olds falling from 49.7% to 31.3% after commencement, platform-level retention of 63.6% to 69.4% among children who had accounts on Facebook, Instagram, Snapchat, and TikTok (48.5% for YouTube), with the most common reason for retention being that the platform had not yet asked the child to verify (66.8%). The same document reports no discernible change in cyberbullying and image-based-abuse complaints in January and February 2026 against the prior year, alongside about 4.7 million under-16 accounts actioned in the first weeks; eSafety's own caveat is that the figure counts accounts, never users. Critics who quote a single "70% still have access" number are compressing that parent-survey record past what it says; the per-platform, parent-reported basis belongs next to the figure.

What the withdrawn channel carries is documented in Canadian sources the federal record never engages. Inuit Tapiriit Kanatami's National Inuit Suicide Prevention Strategy names social media as both a risk and a suicide-prevention opportunity for Inuit youth and commits ITK to researching it as connective infrastructure; First Nations-partnered research in the Canadian Journal of Communication found social networking supports community resilience and cultural preservation in some of Canada's most isolated First Nations; We Matter, the Indigenous youth-led life-promotion organization, delivers its campaign to Indigenous youth through the platforms the restriction would close to under-16s. The other direction is also Indigenous-voiced: the Native Women's Association of Canada's parliamentary brief identifies online recruitment of Indigenous girls as a growing trafficking vector and asks for internet-safety programming.

The bill's record on these communities is one sentence. The release and backgrounder credit "engagement with victims and survivors, civil society organizations, Indigenous partners, experts, industry and Canadians"; the government's own consultation page documents that the Indigenous engagement was a contractor-run sharing circle and interviews with 25 participants in 2022-23, producing a non-attributable report that remains unpublished. As of publication, none of the national Indigenous organizations (AFN, ITK, MNC, the Caring Society, the Friendship Centres) has published anything on C-34, C-63, or online-harms legislation in either direction; the bill is a day old, and that silence is recorded here as a finding about the record. The structural mitigations inside the bill are real: private messaging is excluded (s. 11), so one-to-one and small-group channels survive, and the s. 29 exemption can re-admit under-16s to services that prove adequate safeguards. The implication: the size of the belonging loss depends almost entirely on how fast the exemption pathway operates, and the bill does not sequence it ahead of the restriction.

Ethical and governance implications

An age restriction for some is an age check for everyone. Geist states the structural point: "an age-based restriction is, in practice, an age-verification requirement for everyone, since identifying who is under 16 means identifying everyone who is not" (June 10 analysis); OpenMedia's same-day release warns of "a face scan or ID requirement for dozens of services." The accuracy record at the 16 threshold is mixed by instrument: NIST's evaluation puts the best facial age-estimation error near 2.4 to 3.1 years with errors almost always higher for female faces; Australia's trial concluded age assurance "can be done" privately and effectively, while an adversarial re-analysis of the trial's own data found systems passing 43% to 45% of 10-year-olds through a 16+ gate and one system passing 84% of tested First Nations children under 16. The Internet Society's committee submission adds the exclusion case: Canadians without government ID or bank accounts get marginalized by verification regimes. Against this, the bill's own conditions are stricter than its critics' summary of it: measures qualify only if effective, purpose-limited, and destroyed after the check (ss. 22(2), 27(2)); the Commission must consult the Privacy Commissioner before the key regulations (s. 122(2)); and the OPC published age-assurance guidance five weeks before the bill, still in open consultation until August 4, 2026.

The Commission concentrates rule-making, investigation, and enforcement in one appointed body, and the bill names no appeal route from its decisions; the Statutory Instruments Act is disapplied for several of its order types. The CCLA's standing position from C-63 is that such concentration "undermines the fundamental principle of democratic accountability"; the Canadian Constitution Foundation's Christine Van Geyn predicted over-removal of "controversial, transgressive, offensive and unpopular speech" (Canadian Press). The text answers partway: the duties cannot require measures that unreasonably or disproportionately limit users' expression (ss. 22(3), 27(3), 32(3)), and s. 12 disclaims any general content-monitoring obligation. Two record-keeping defects belong on the page: the release's quick facts and its own backgrounder disagree on whether user complaints reach the Commission under one duty or all three; and the release quotes Statistics Canada's 16,905 online child sexual exploitation incidents with the 347% ten-year rise while omitting that the same StatCan release reports a 16% rate decline from 2023.

The chatbot provisions are the bill's most novel ethics. Section 51's triad (immediate interruption, referral to services available at that moment, interaction with a human being) exists in no other enacted law: California's SB 243 and New York's Article 47 both stop at a referral notification. The duty's endpoint is evidence-based: 88.1% of surveyed suicidal Lifeline callers said the call stopped them from killing themselves, and a 2026 JAMA analysis linked 988 expansion to an 11% drop in youth suicide deaths. The protocol layer the duty regulates is unproven: the RAND audit found ChatGPT, Claude, and Gemini inconsistent at intermediate risk; no deployed system interrupts, checks live availability, and hands to a human; and clinical commentary warns that abrupt hard-stops can themselves harm users in crisis. Geist reads the definition as "a capability test, not a purpose test" that sweeps in every general-purpose assistant; the definition's text (an AI system "capable of being used... to simulate a sustained human-like relationship") is on the page for the reader to judge. The implication: s. 51 mandates the strongest crisis duty in the world on a mechanism with no outcome evidence, pointed at an endpoint with strong outcome evidence.

Where Bill C-34 sits on the quadrant

The Transmutarianism framework scores agents on F (filtering of deprivation: deprivation absorbed without being passed on) and A (amplification of fulfillment: fulfillment emitted in excess of what was received). Moral work M = [τF + A] / √(τ²+1) (at the parity origin F₀ = A₀ = 0) is computed per Maslow level (physiological, safety, belonging, esteem, actualization) and weighted by w = {5, 4, 3, 2, 1}. The normalizer keeps M comparable across τ settings; a power coefficient ρ can raise the bar for powerful agents.

The chart below plots Bill C-34's provisional placement (sigil-red outline) against the four archetype reference dots (muted, shown for orientation). Drag the τ slider to test sensitivity. Click any dot for its F, A, and M values.

Assumptions, stated

1. Scope of evaluation. Bill C-34 as tabled at first reading on June 10, 2026: the Digital Safety Act and the Digital Safety Commission of Canada Act it would enact, scored as the regime the text constructs. Excluded: the hate-speech enforcement components C-34 drops from C-63 (Criminal Code, Canadian Human Rights Act), the content of future Governor in Council regulations beyond what the bill's text constrains, and the conduct of individual platforms. One inside-scope design split is material to the character of the bill and is scored as found: social media services face an age restriction while chatbot services face duties without any age gate.
2. Time horizon. Steady state at full implementation: royal assent, order-in-council commencement, services designated, Commission operational, which the sponsor's own timeline places no earlier than roughly 2028. Implementation has three phases on that timeline: pre-commencement (today until the order in council), where the bill emits no operative flows and the dot sits at the Conduit origin; a commencement-to-Commission window (designation at six to eight months against a Commission at 18 months), where the under-16 restriction can be in force with no regulator and no s. 29 exemption pathway; and the funded steady state scored here. The horizon sensitivity below reports both non-steady-state dots.
3. Asymmetry coefficient (τ). Default τ = 1. Sensitivity at τ = 0.8 (flourishing focus) and τ = 1.5 (cycle-breaking focus) is shown in the math box below.
4. Maslow weighting. w = {5, 4, 3, 2, 1} for {physiological, safety, belonging, esteem, actualization}; lower-level deprivation is heavier. The flat-weight recompute is in the sensitivity set.
5. F and A scale. Each level scored on a −10 to +10 range, central estimate from the public-record evidence above, substitutable; the reasoning column names the inputs.
6. Denominator. Canadian children and youth under 18: the population the bill's duties are built around ("child" means under 18 throughout the Digital Safety Act, with the under-16 restriction a subset). This is an active choice: the bill's object and its dominant flow path both run through this population. Excluded flows: platform shareholders, the age-assurance vendor market, and adults' flows except where the bill routes them through children. The strongest alternative denominator a critic would choose is all Canadian users of regulated services, since the under-16 restriction operationally requires age-checking everyone; the alternatives table below recomputes the dot under it.
7. Treatment of diverted resource. Excluded as counterfactual in the central table. The diverted public resource is the Commission's operating cost, which no current document prices: the honest central is uncosted, inside a band from roughly $10 million a year (the run rate implied by Budget 2024's $52 million over five years) to $57 million a year (the PBO's three-body C-63 proxy, about $1.40 per Canadian per year). The bill consumes no new land, energy, or housing. Cost-recovery charges under s. 125 would shift the incidence to operators and, per the fiscal section's passthrough sentence, partly back to users. The alternative treatment (scored as a negative physiological flow) is recomputed below at the band's upper anchor, with the doubled-magnitude check stated, and the composed construction (all-users denominator with the diversion scored where its incidence falls) appears in the alternatives table.
8. Power-asymmetry (ρ and Δ_M). Central verdict at ρ = 1 (parity), as a default awaiting calibration. The operational power trigger holds: the subject is the more powerful party in the audited flow and controls the regulatory resource the denominator population depends on, so the Δ_M band is reported below, in per-level F/A units (with a uniform per-level handicap, W falls by 15·Δ_M).

Per-Maslow scoring

The aggregate dot above is computed from the per-level table below. Each F and A value is the central estimate from the public-record evidence in the prior sections; substitute different numbers and the dot moves. The per-level scores were derived blind, before any prior field note's numbers were consulted, and recorded in the note's assumption-challenge ledger; the ledger, including the full challenge matrix from the pre-publication gate, is available on request via sev@economyofwisdom.com.

Where the dot lands under the alternatives

These are the alternative constructions from the pre-publication assumption gate, recomputed in full.

The quadrant is stable: every tested construction lands Absorber. The sign and size of W are less stable, and the pre-publication gate surfaced three places where the sign turns. First, the commencement-to-Commission window scores W = −7.8 under the standing construction above; the dissent reading, which spreads voluntary statutory compliance evenly across the duties instead of concentrating it on the restriction, keeps W positive at any uniform compliance fraction. Second, the composed all-users construction sits at W = +0.4 and falls to −1.4 when platform compliance costs pass through at the UK-benchmarked magnitude; doubling the central diversion magnitude alone moves the diversion-scored row only to 14.1. Third, the power handicap flips the sign of W at Δ_M ≈ 1.18 on the central denominator and at Δ_M ≈ 0.14 under all-users. Alternative defensible score sets from the same evidence place W between 4.9 and 19.8, Absorber throughout; a regulation-downside scenario (narrow designation, diluted destruction conditions) gives 12.0; a multiplicative-mirror τ band ({0.67, 1.0, 1.5}) crosses zero only on the all-users row, just below the prescribed band's lower endpoint. Whether C-34 is a strongly positive instrument, a barely positive one, or in its worst window and harshest composed reading a slightly negative one depends on whose flows a reader counts, over which window, against what power baseline; those are values questions the framework surfaces rather than settles. The Foundation will publish a re-audited placement if the missing instruments (funding, regulations, Charter Statement) materially change any cell.

Public-interest recommendations

Committee study is where a first-reading text either acquires its missing instruments or ships without them.

For Parliament and Canadian Heritage

1. Table the Charter Statement and a PBO-grade costing with a funding instrument before second-reading votes; a regulator with duties and no budget line filters nothing.
2. Sequence the s. 29 exemption criteria so they exist before the under-16 restriction takes effect for any service; the current officials' timeline runs the restriction ahead of the exemption.
3. Resolve the complaints-scope inconsistency between the release (one duty) and the backgrounder (all three duties) in the text itself.
4. Add the statutory duty to consult young people that UNICEF Canada has requested, with a child rights impact assessment.
5. Publish the 2026 Expert Advisory Group session summaries and the 2022-23 Indigenous engagement report before committee study closes; engage the national Indigenous organizations directly, since remote-community connectedness is a named stake in the evidence.

For the Digital Safety Commission of Canada, once constituted

1. Set published accuracy and bias floors for age-assurance instruments at the 16 threshold (error by age band, sex, and skin type, NIST-style), and reject instruments below them; the Australian trial's own data shows the failure modes.
2. Require s. 51 implementations to be warm handoffs rather than hard terminations, and require operators to report referral counts and completion rates; the clinical literature warns that abrupt cut-offs can harm the user they interrupt.
3. Use the s. 129 three-year review as an outcome evaluation with pre-registered metrics (victimization rates, connection measures, complaint volumes), on the model of Australia's two-year evaluation with an academic advisory group.

For provinces and school boards

1. Keep classroom-level and litigation-discovery data flowing as independent measurement; 2025 ONSC 1499 makes the school-board record a discovery channel no federal regulator controls.

For platforms and AI chatbot operators

1. Publish crisis-referral outcome data (trigger counts, click-throughs, completion) ahead of any mandate; no operator publishes any today, and the first mandated disclosure anywhere (California's) is not due until July 2027.
2. Apply for the s. 29 exemption pathway with audited safeguards rather than lobbying against the restriction; the exemption is the mechanism that converts the ban into conditional market access.

The audit protocol (six inputs)

An audited placement requires six inputs. The Foundation offers framework, scoring template, methodological support, and the published report to any party willing to supply them. For this subject, the record currently lacks several of them; the missing-document register in the ledger names each.

1. The funding instrument and approved budget for the Digital Safety Commission of Canada, by fiscal year.
2. The designation regulations: user-number thresholds, the services specified for the under-16 restriction, and the rationale for each.
3. Age-assurance audit data at the 16 threshold: error rates by age band, sex, and skin type; data-retention and destruction compliance.
4. Platform transparency data: under-16 accounts actioned, complaint volumes and outcomes, takedown timing against the 24-hour duty.
5. Chatbot crisis-protocol data: s. 51 trigger counts, referral completion rates, human-handoff rates, and false-positive interruption rates.
6. The s. 129 review's outcome metrics: youth victimization, mental-health, and connection measures, pre-registered before commencement.

Contact: sev@economyofwisdom.com.

What changes the placement

Toward Transmuter (A turns positive): the s. 29 exemption pathway operating before the restriction bites, so safe-by-design services re-admit under-16s and the ban functions as conditional market access; a statutory youth-consultation duty and child rights impact assessment in the text; funded connection alternatives for remote and Indigenous communities named in the implementation plan; a s. 129 review that measures connection alongside victimization and finds the protections delivering without the withdrawal.

Toward Extractor (F turns negative): age assurance implemented as ID or biometric collection that is retained, breached, or repurposed, since the data infrastructure outlives any ban it serves; the Commission operating unfunded, leaving duties on paper while compliance costs pass through to users; over-removal of lawful youth expression at the scale the CCF predicts; s. 51 implemented as hard termination that deters disclosure from users in crisis.

Better data moves the dot.

A note on framing

The Foundation publishes these notes to give the parties at the table a shared instrument: one that is harder on the evidence than a press release and more honest about its own assumptions than an advocacy campaign. Every number above can be disputed by substituting a different number with a source, and the dot will move. That is the design.

Bill C-34 sits where caregivers often sit on this map: absorbing harm on behalf of someone smaller, and paying for it out of the relationship itself. This Absorber's moral work is positive (W = 17.7 at parity); the framework's caution about Absorbers is that filtering without emission runs on a reserve that depletes. An Absorber protects; a Transmuter also gives. The version of this bill that crosses into Transmuter territory is visible from here, and it is mostly a sequencing question: let the exemption pathway, the consultation duty, and the funded alternatives arrive with the restriction instead of after it.

The classification is information about flows. The people inside Canadian Heritage, the future Commission, and the platforms are working in incentive structures the framework's purpose is to make visible; the recommendations are addressed to the institutions. The math is at transmutarianism.org/framework/. The dot is plotted on the live quadrant explorer at transmutarianism.org/quadrant/. To dispute the placement, substitute different F and A values per level, with a source for each, and the dot moves.