The guide has documented specific Canadian AI policy gaps across multiple domains: training data consent and compensation (Chapter 11), environmental…

The guide has documented specific Canadian AI policy gaps across multiple domains: training data consent and compensation (Chapter 11), environmental disclosure and verification (Chapter 8, Chapter 18), Indigenous data sovereignty integration (Chapter 9), workplace AI surveillance limits (Chapter 10, Chapter 16), comprehensive privacy regulation (Chapter 10), political deepfake coverage (Chapter 13), algorithmic decision transparency (Chapter 10, Chapter 15), AI infrastructure sovereignty at the operational rather than rhetorical level (Chapter 9), and the broader transparency-as-precondition gap (Chapter 18) that underlies most of the others. This chapter asks the constructive question: what specific policy mechanisms could close these gaps, through what institutional pathways, with what political coalitions, with what trade-offs that any honest analysis has to engage? The chapter is constructive but not utopian. Every mechanism examined here has real trade-offs, political constraints that have prevented action so far, and a supporting coalition whose mirror image is the opposition it would generate. The chapter does not argue that any specific mechanism is obviously correct; it argues that specific mechanisms exist, that the pathways to implement them have been mapped, and that the political conditions for activating each pathway are knowable even when they are not currently aligned. You will leave the chapter with: the highest-leverage convergent policy mechanisms named with their cross-chapter benefits; specific federal legislative mechanisms with the trade-offs honestly engaged; the federal-provincial coordination pathway and what would be required to activate it; the sectoral regulator empowerment pathway with specific institutional examples; the international engagement pathway including the Council of Europe Framework Convention question; the Indigenous data sovereignty integration mechanism that AI for All did not commit to; and the working understanding that closing the gaps is possible but politically constrained in ways the guide has documented. ---

The convergent high-leverage mechanisms

The guide's analysis across multiple chapters has surfaced several mechanisms that would address gaps in more than one policy domain simultaneously. These convergent mechanisms are the highest-leverage interventions — the policy changes that pay off across multiple chapters rather than just one. Foregrounding them is the chapter's most useful constructive contribution.

Mechanism 1: Comprehensive likeness rights in Canadian copyright.

The guide identified this convergence in Chapter 11 (IP & Copyright) and Chapter 13 (Misinformation & Deepfakes): ACTRA's NIL Rights argument and CIGI's likeness-rights-in-copyright argument converge on a single mechanism — a recognized rights claim, in Canadian law, over a person's likeness, name, image, and voice (defined in full in Chapter 11).

What it would address: - Performer NIL rights in commercial contexts (Chapter 11) - Non-consensual deepfakes across categories: sexual, political, commercial, identity-fraud (Chapter 13) - Specific recourse for ordinary Canadians against unauthorized likeness use - A civil remedy faster than criminal prosecution under Bill C-16 - Coverage of political deepfakes that current Canadian law does not address

The institutional pathway: Federal legislation amending the Copyright Act, or new standalone legislation establishing likeness rights as a distinct statutory category. Denmark's Copyright Act amendment (introduced 2025; pending final adoption as of mid-2026, with entry into force planned for 2026) provides the most developed legislative precedent, including the satire and parody exceptions a Canadian version would need.

The political coalition: Creator organizations (ACTRA, the Writers' Union of Canada, the Directors Guild of Canada, Music Canada) plus civil liberties organizations (the Canadian Civil Liberties Association, the BC Civil Liberties Association) plus privacy advocates (the Office of the Privacy Commissioner has expressed interest in likeness-rights extensions). The convergence across the political spectrum is unusual and significant.

The trade-offs that have to be engaged: - Free expression concerns about restrictions on synthetic content used in satire, education, and journalism. These are real and not dismissible. The Danish model includes specific exceptions for these purposes; the Canadian implementation would need to engage them carefully. - Definitional complexity around what constitutes "likeness" — particularly for voice cloning, stylistic imitation, and AI outputs that resemble specific individuals without exactly reproducing them. - Enforcement complexity for content hosted on foreign platforms outside Canadian jurisdiction. - Potential overlap with existing common-law misappropriation of personality doctrines, which would need to be statutorily clarified.

The honest assessment: Likeness rights legislation is the single highest-leverage Canadian AI policy mechanism currently visible. It addresses gaps in two chapters' worth of documented harm with a single legislative vehicle. The political coalition exists across multiple existing constituencies. The trade-offs are real but not insurmountable. The current government has not committed to this mechanism; whether a future Parliament will commit is contested. The mechanism remains available regardless.

🧌 GOBLIN CHECK — When performers' unions and privacy researchers walk into Parliament asking for the same statute for opposite reasons, that is not a coincidence. That is a coalition lying on the ground, waiting for someone to pick it up. The goblin files this under "rare," cross-referenced under "so why is nobody picking it up."

Mechanism 2: Mandatory comprehensive AI transparency framework.

Chapter 18 established that transparency is the operational precondition for most other AI policy work. This mechanism would provide the disclosure infrastructure that copyright enforcement, environmental policy, bias audit, worker protection, and sovereignty analysis all require — and each of those downstream uses maps to a specific disclosure: training-data composition (copyright determination and bias audit), environmental impact including verified third-party measurement (environmental policy), algorithmic decision transparency (worker and consumer protection), supply-chain transparency (sovereignty analysis), and AI deployment disclosure (accountability across sectors).

The institutional pathways are multiple:

Federal legislation: comprehensive AI transparency requirements analogous to the EU AI Act's disclosure provisions, applicable to AI providers placing systems on the Canadian market regardless of corporate jurisdiction. Substantial federal capacity exists through the Office of the Privacy Commissioner, the Treasury Board, and existing federal regulators.

Federal-provincial coordination: agreements among Ottawa and provinces establishing consistent transparency requirements across jurisdictions, similar to existing federal-provincial coordination on climate policy and labour-market matters.

Sectoral regulator expansion: empowering existing regulators (OPC, CRTC, OSFI, Health Canada, Competition Bureau) to require AI transparency within their sectoral mandates. This pathway is faster than legislation but produces fragmented coverage.

International alignment: adopting EU AI Act-equivalent provisions to enable Canadian providers operating internationally to comply with a single framework, and to maintain market access for Canadian AI exports to the EU market.

The political coalition: Civil society advocacy organizations across multiple sectors (CUPE, the Canadian Civil Liberties Association, the Writers' Union of Canada, ACTRA, the Federation of Canadian Municipalities, Indigenous data sovereignty organizations, environmental NGOs), academic researchers, journalism organizations, and (perhaps unexpectedly) some segments of the AI industry that face significant compliance complexity from inconsistent international frameworks.

The opposition: Substantial. Industry advocacy against comprehensive transparency frameworks has been consistent globally — the documented pattern of declining transparency in the Stanford FMTI (Chapter 18) is consistent with several explanations — litigation exposure, competitive secrecy as models scale, or deliberate strategic positioning — and the Index documents the trend without pinning down which. Read it as a question, not a verdict. The political-economy of comprehensive AI transparency requires sustained government commitment against organized industry resistance.

The assessment: Comprehensive AI transparency legislation is the most significant policy mechanism currently visible. It addresses gaps across virtually every other chapter of this guide. The political coalition for it exists but is currently dispersed across multiple advocacy organizations without coordinated parliamentary advocacy. The opposition is well-resourced and concentrated. Whether the political conditions emerge for comprehensive legislation depends substantially on coalition coordination that has not yet happened at the scale required.

Mechanism 3: Indigenous data sovereignty integration in federally-funded AI.

The guide has documented across Chapter 9 and Chapter 15 that the federal AI for All strategy does not commit to integrating OCAP® (First Nations), NISR (Inuit), or CARE (international Indigenous) data sovereignty frameworks into federally-funded AI work. This mechanism would address that gap. It would establish Indigenous data sovereignty over training data containing Indigenous content, Indigenous governance over federally-funded research involving Indigenous communities, Indigenous procurement criteria for federal AI contracts, and specific protections for Indigenous languages, knowledge systems, and cultural content.

The pathway here does not require legislation. A Treasury Board directive could require all federally-funded AI work to engage applicable Indigenous data sovereignty frameworks, drawing on existing FNIGC OCAP-certification infrastructure, ITK research-protocol mechanisms, and the Abundant Intelligences research network (already federally funded at more than $22M through the New Frontiers of Research Fund). The coalition is well-organized: Indigenous-led organizations (Assembly of First Nations, Inuit Tapiriit Kanatami, Métis National Council), data sovereignty research organizations (FNIGC, the Abundant Intelligences program, the Global Indigenous Data Alliance), Indigenous AI researchers, and civil society supporting Indigenous self-determination broadly.

The trade-offs are mostly definitional and jurisdictional. What counts as Indigenous data (data about Indigenous communities versus data generated by Indigenous individuals) is contested. Applications that span multiple communities with different governance protocols are hard to standardize. A federal directive has to be reconciled with the legitimate authority of individual communities to set their own terms. And the deepest question (whether non-Indigenous-led federal directives can adequately operationalize Indigenous data sovereignty at all) is one the Indigenous-led organizations would need to answer through their own deliberative processes.

The honest assessment: this is achievable through existing institutional mechanisms without new legislation. The federal government already funds the Abundant Intelligences program; integrating its frameworks into the broader strategy would not require building new infrastructure. The coalition exists. The barrier is political will at the cabinet level, which AI for All did not exhibit, and whether a future cabinet does is contested.

---

Federal legislative mechanisms — what comprehensive AI law could include

Beyond the convergent mechanisms above, several specific federal legislative mechanisms would address documented gaps. The chapter engages each with honest attention to trade-offs.

A successor to AIDA. Chapter 17 documented AIDA's failure and the current government's choice not to revive it. A successor framework would need to address the specific criticisms that caused AIDA's failure: inadequate consultation, vague definitions, contested enforcement model, packaging complexity. The institutional path exists; the political will has not emerged.

What a successor framework could include, addressing the specific gaps the guide has documented:

Risk-based classification of AI systems (drawing on the EU AI Act framework while adapting to Canadian federalist constraints)
Mandatory algorithmic impact assessments for high-impact AI deployment
Specific provisions for AI used in employment decisions, healthcare, financial services, and consumer protection
Specific provisions for AI used by law enforcement, including biometric identification limits
Comprehensive transparency requirements paralleling the convergent mechanism above
A federal AI commissioner with cross-cutting authority and substantial enforcement powers
Coordination provisions with provincial regulators
Specific provisions for Indigenous data sovereignty integration

The political coalition for a successor framework would need to be substantially broader than AIDA's. The criticisms of AIDA included inadequate consultation with civil society, Indigenous communities, and labour organizations; a successor would need to engage all three substantively before introduction. The criticisms also included industry concerns about regulatory burden; a successor would need to engage industry technical input without ceding the comprehensiveness that civil society demanded.

Where this leaves things: A comprehensive federal AI law could be drafted that addresses AIDA's specific failures while maintaining the comprehensive scope. The drafting is possible. The political conditions for introduction, parliamentary support, and passage have not been assembled. This is the largest unfilled policy commitment in Canadian AI governance.

PIPEDA modernization with AI-specific provisions. Privacy Commissioner Philippe Dufresne has publicly called for substantial PIPEDA reform including AI-specific provisions. The reform would address:

Inferential extraction (the AI-distinctive privacy problem from Chapter 10)
Algorithmic decision rights (notification, explanation, appeal — Quebec Law 25-equivalent at the federal level)
Mandatory privacy impact assessments for high-risk AI
Significant penalties for non-compliance (GDPR-equivalent scale)

The political coalition: The Privacy Commissioner has been active. Civil society support exists. Industry support for some elements (regulatory clarity, alignment with international frameworks) and opposition for others (mandatory PIAs, significant penalties).

The trade-off engagement: PIPEDA modernization has been technically ready for legislative action for several years. The legislative vehicle that would carry it (a successor to Bill C-27's CPPA component) has not been introduced by the Carney government. The barrier is political prioritization, not technical readiness.

Comprehensive AI environmental disclosure legislation. Chapter 8 documented the gap between corporate AI environmental claims and operational reality. The Luccioni/Strubell/Crawford Jevons' paradox analysis provides the framework for understanding why per-unit efficiency claims don't substitute for aggregate measurement.

Federal legislation could require: - Mandatory environmental impact reporting for AI providers placing systems on the Canadian market - Verified third-party measurement methodology - Lifecycle assessment including embodied carbon, water consumption, and end-of-life environmental impact - Specific requirements for data centres above thresholds (paralleling the EU data-centre reporting scheme) - Sasha Luccioni's proposed Energy Star-equivalent labelling for AI systems

A specific operational threshold worth foregrounding. The EU's data-centre reporting scheme sets the mandatory threshold at 500 kW of installed IT power. A Canadian equivalent could reasonably set the threshold at 5-10 MW, substantially higher than the EU baseline, capturing only the larger facilities, but still bringing the major AI-relevant operators into mandatory disclosure. Required reporting could include: annual electricity use, peak demand, PUE, direct water consumption with source mix, backup generation emissions, and demand-response or curtailment capability. This is the specific mechanism that would close the no-Canadian-facility-level-census gap documented in Chapter 18. The threshold choice is a calibration question (set it too low and the regulatory burden expands without proportionate benefit, set it too high and major facilities escape disclosure), but the underlying mechanism is well-established internationally.

GOBLIN FACTS — the threshold already exists; the question is whether Canada copies it. The EU requires every data centre with 500 kilowatts or more of installed IT power to report its energy and water use every year. Canada requires none of this federally, and only about 22% of Canadian data centres report their efficiency at all. The lever is built and in use one ocean over. Nobody in Ottawa has reached for it.

The provincial allocation regimes documented in Chapter 6 are themselves operational precedents for this kind of disclosure infrastructure. Ontario's IESO already requires substantial information from prospective large-load connections; Alberta's AESO collected the data underlying its 1,200 MW interim cap and 16 GW queue documentation; BC Hydro's competitive allocation framework requires AI projects to compete based on documented operational characteristics; Hydro-Québec's proposed large-data-centre tariff implies the same information collection. The disclosure infrastructure that mandatory federal reporting would establish is, in significant part, already being collected by provincial grid operators. It just isn't being shared with the public or with federal regulators. A federal reporting standard could draw on the data the provincial operators already require, with disclosure obligations layered on top. The political coalition for this would include the provincial grid operators themselves, who have already established the basic data-collection infrastructure.

The political coalition: Environmental organizations, civil society broadly, academic researchers (Luccioni's network is Canadian-anchored at Hugging Face Montréal), labour organizations concerned with broader environmental impact, provincial grid operators (Ontario IESO, Alberta AESO, BC Hydro, Hydro-Québec) whose existing operational data collection would be substantially leveraged by federal disclosure standards. Opposition: AI industry positioning against measurement frameworks that don't flatter results.

The assessment: AI environmental disclosure legislation is feasible and would address one of the guide's most-documented gaps. The political coalition exists. The provincial allocation precedents make the case stronger than a year ago. The current government has not committed to it.

---

The federal-provincial coordination pathway

Chapter 17 documented the federalist constraint that limits direct federal authority over much of the AI deployment that would benefit from comprehensive regulation. The federal-provincial coordination pathway is the mechanism for achieving comprehensive coverage despite the constitutional structure.

The historical precedents. Federal-provincial coordination has produced comprehensive frameworks in multiple policy domains — climate policy (the Pan-Canadian Framework on Clean Growth and Climate Change, with substantial provincial variation), labour-market policy (the Workforce Development Agreements), pharmaceutical pricing (the pan-Canadian Pharmaceutical Alliance), housing policy (the National Housing Strategy with provincial bilateral agreements). The pathway is established; AI policy could follow similar institutional logic.

EXAMPLE — Canada already builds big things this way. Ottawa can't simply order the provinces around on health, labour, or education, so on climate, drug pricing, and housing it negotiates a shared framework instead. AI governance could ride the same rails. The machinery for doing hard things across jurisdictions exists and gets used. It just hasn't been pointed at AI yet.

What federal-provincial AI coordination could include:

Consistent privacy frameworks across provinces. Right now the coverage is a patchwork. Quebec's Law 25 is comprehensive. PIPEDA covers the federal private sector. Provincial private-sector laws in BC, Alberta, and Manitoba cover part of the field, and the remaining provinces fall back on PIPEDA. A federal-provincial agreement could establish consistent privacy protections, including AI-specific provisions, across all jurisdictions.

Consistent workplace AI regulations. Currently Ontario's Working for Workers Act provides limited disclosure requirements for hiring AI; Quebec's Law 25 provides automated decision rights; most other provinces have minimal protection. A federal-provincial agreement could establish baseline workplace AI protections.

Consistent healthcare AI frameworks. Healthcare is primarily provincial jurisdiction. Federal-provincial coordination could establish consistent standards for AI in clinical decision support, diagnostic imaging, hospital administration, and public health applications.

Consistent education AI frameworks. Education is provincial. Coordination could address AI in admissions, grading, learning analytics, and student data governance.

The political coalition: Federal government leadership (which has been absent on this pathway), provincial government cooperation (which varies substantially by province and political composition), civil society organizations operating across jurisdictions, and academic researchers documenting the inconsistency.

The trade-offs that have to be engaged:

Provincial autonomy concerns. Some provinces (Quebec most prominently, but also Alberta) have political traditions of resisting federal coordination perceived as encroaching on provincial jurisdiction. Coordination agreements need to be negotiated rather than imposed.

Speed-vs-comprehensiveness trade-off. Provincial-led action (like Quebec's Law 25) can be faster than federal-provincial coordination, but produces inconsistent coverage. Federal-provincial coordination produces comprehensive coverage but takes longer.

Implementation complexity. Even when agreements are reached, implementation varies across provinces with different administrative capacities and political priorities. Climate policy has demonstrated this — the Pan-Canadian Framework was reached but implementation has varied dramatically.

The honest assessment: Federal-provincial coordination is the institutional pathway most capable of producing comprehensive Canadian AI policy that addresses the federalist constraint. It is also the pathway with the longest timeline and the most political complexity. The pathway has not been activated by the current federal government. Whether it will be is contested. The pathway remains available regardless of current political conditions.

---

Sectoral regulator empowerment

The faster but more fragmented pathway is empowering existing federal and provincial regulators to develop AI-specific frameworks within their sectoral mandates. Chapter 17 documented that this is Canada's de facto current pattern. Substantial expansion within this pathway would address many specific gaps without requiring new comprehensive legislation.

The OPC. Privacy Commissioner authority could be expanded through PIPEDA reform and through provincial commissioner coordination. Substantial AI-specific enforcement is already being developed under existing authority; statutory clarification would enable systematic rather than case-by-case engagement.

The CRTC. Telecommunications and broadcasting regulator authority could be extended to AI deployment in regulated sectors — AI in broadcast content, AI in telecommunications customer interactions, AI in content moderation for regulated platforms.

OSFI. Financial regulator authority extends to AI deployment in federally-regulated financial institutions. Substantial AI guidance is being developed; statutory enhancement could enable more systematic enforcement.

Health Canada. Medical device AI regulation is being developed through existing Health Canada authority. Provincial healthcare AI regulation could be coordinated through Health Canada frameworks.

The Competition Bureau. AI market concentration concerns could be engaged under existing Competition Act provisions, with possible statutory enhancement to address AI-specific competition issues.

The Auditor General. Federal AI deployment audit could be expanded substantially without requiring new statutory authority.

Provincial commissioners and regulators. Provincial action within existing jurisdiction can produce significant coverage. Quebec, BC, Alberta, Manitoba, and Ontario have all moved on specific AI matters; expansion within existing provincial authority is feasible.

The assessment: Sectoral regulator empowerment is the pathway most accessible without new legislation. It would produce fragmented coverage but could address many specific gaps relatively quickly. The pathway has been partially activated; substantial additional activation is possible within existing institutional authority. The barrier is political will at the responsible-minister level for each regulator, and resource allocation for enforcement.

---

International engagement

Canada's position in international AI governance was documented in Chapter 17. The international pathway includes specific mechanisms that would close documented gaps:

Council of Europe Framework Convention on AI. Canada signed this first binding international AI treaty on February 11, 2025, but has not ratified it. Ratification plus implementing legislation would convert the signature into domestic legal commitments. The barrier, as elsewhere, is legislative prioritization rather than eligibility. (The EU became the treaty's first ratifying party in May 2026; the treaty is not yet in force.)

Beijing Treaty on Audiovisual Performances. Canada has neither signed nor acceded to this 2012 WIPO treaty. Accession plus implementing legislation would extend moral rights to audiovisual performers (relevant to the ACTRA NIL Rights advocacy from Chapter 11). Accession is procedurally straightforward; the political decision and the implementing legislation have not occurred.

EU AI Act alignment. Without joining the EU AI Act framework directly (which Canada cannot do as a non-EU member), adopting equivalent domestic provisions would simplify compliance for Canadian companies operating in the EU market and would establish operational interoperability.

Sovereign Technology Alliance development. The Alliance (launched with Germany in February 2026, folded into AI for All alongside twelve signed international AI partnerships) could become a substantive international coordination mechanism if developed seriously. Currently the Alliance's operational framework is not fully announced. The development would require sustained Canadian leadership.

UN engagement. Canadian participation in UN AI processes could be more substantive. The UN Secretary-General's Advisory Body on AI, the UNESCO AI Ethics Recommendation, and the broader UN framework provide pathways for international policy development that Canada participates in but does not lead.

Where this leaves things: International engagement is the pathway most easily activated without requiring controversial domestic political conditions. Signing the Council of Europe Framework Convention, ratifying the Beijing Treaty, and developing the Sovereign Technology Alliance substantively are all straightforward government decisions. The barrier is political prioritization within the cabinet's broader agenda, not external constraints.

---

Judicial pathways and what they can and cannot deliver

The Canadian newspapers v. OpenAI case (Chapter 11) is the most-significant active Canadian AI litigation. The case will produce precedent. Multiple other cases are working through the courts on AI-related copyright, privacy, employment, and consumer-protection issues. The judicial pathway is operational; the question is what it can and cannot deliver.

What the judicial pathway can deliver: - Specific precedent on contested legal questions - Authoritative interpretation of existing statutes as they apply to AI - Damages and remedies for documented harms in specific cases - Pressure on legislators to clarify statutory frameworks - Public information through published rulings and trial proceedings

What the judicial pathway cannot deliver: - Systematic regulation of AI deployment (courts decide cases brought before them, not comprehensive policy) - Disclosure infrastructure that requires statutory authority - Resources for affected parties to bring cases (litigation is expensive and slow) - Coverage for harms that don't reach litigation thresholds

The strategic question for civil society and government: which gaps should be pursued through legislation, which through litigation, and which through some combination. The Canadian newspapers v. OpenAI case is being pursued because the plaintiffs have resources and standing; many similar cases will not be pursued because affected parties lack resources. Litigation can produce important precedent but cannot substitute for systematic policy.

---

The political-economy conditions

The chapter has mapped multiple pathways. The honest engagement with what could close the documented gaps requires also engaging the political conditions that have prevented action so far.

The current Canadian political moment. The Carney government took office in 2025 and committed to AI for All as the federal AI strategy, investment-focused rather than regulation-focused. It has not committed to reviving AIDA, nor to comprehensive privacy reform — and the same blank runs across the rest of the guide's documented gaps: no commitment to expanding the Treasury Board AI Register, to Indigenous data sovereignty integration, to AI environmental disclosure, to political deepfake legislation, to federal-provincial coordination, or to ratifying the Council of Europe Framework Convention it signed in February 2025. The current government's choice is to address AI primarily through industrial strategy rather than through regulation. This is a choice; other choices are available.

The opposition political conditions. The Conservative party and other opposition parties have not advanced comprehensive AI policy alternatives. The opposition framework is largely focused on contesting specific AI for All commitments rather than on proposing different comprehensive frameworks.

The civil society and advocacy positions. Multiple advocacy organizations have specific proposals — CUPE on workplace AI, the Writers' Union of Canada and ACTRA on creator rights, the BC + AI coalition on sovereignty, Indigenous data sovereignty organizations on integration, the Office of the Privacy Commissioner on PIPEDA reform, the Centre for International Governance Innovation on likeness rights and other matters. The advocacy positions are well-developed; coordinated parliamentary advocacy across positions has not been organized at the scale required to shift cabinet priorities.

The industry positions. AI industry positions are diverse. The major US-headquartered foundation model developers and the Canadian hyperscaler subsidiaries generally oppose comprehensive regulation. Some Canadian creator-industry organizations (publishers, music labels, broadcasters) support specific copyright and licensing legislation. Canadian AI startup positions are mixed.

The labour positions. Major Canadian unions (CUPE, Unifor, the Public Service Alliance of Canada, professional unions for creators) have substantive AI policy positions. The labour positions are aligned on broad principles (worker protection, displacement transition support, regulation of workplace AI surveillance) and varied on specific mechanisms.

The Indigenous positions. Indigenous-led organizations have specific positions on data sovereignty integration. The positions are developed; the cabinet engagement has been limited.

The structural political-economy finding: the gaps the guide has documented have not been closed because the political coalition for closing them has not been assembled and activated at the scale required. The advocacy positions exist. The institutional pathways exist. The barrier is political coordination, not technical or institutional infeasibility. Whether the political coordination emerges depends on factors outside the guide's scope to predict.

---

What this chapter does not commit to

The chapter has presented mechanisms and pathways. It deliberately does not commit to specific recommendations, for reasons that connect to the guide's methodology throughout.

The guide is not a manifesto. It presents what could be done, not what should be done. The "should" questions involve value trade-offs that different reasonable people will weigh differently. The guide's commitment is to making the trade-offs visible rather than to resolving them.

The guide does not endorse specific political coalitions. Different coalitions could support different mechanisms. The chapter has named coalitions that exist; it has not advocated for the formation of any specific coalition.

The guide does not predict political outcomes. The mechanisms exist regardless of whether they are pursued. The pathways are mapped regardless of current political conditions. Whether the political conditions emerge for any specific pathway is contingent on factors the guide cannot predict.

The guide does not claim that closing the gaps would solve AI policy in Canada. Comprehensive AI transparency, likeness rights legislation, Indigenous data sovereignty integration, federal-provincial coordination — even if all of these were enacted, AI policy would remain a contested, evolving domain. The mechanisms address documented gaps; they do not produce comprehensive AI governance once and for all.

The guide does take a position on something: the gaps are real, the institutional pathways for closing them are real, and the political conditions for activating those pathways are knowable. What is done with that information is the reader's responsibility — and the responsibility of the broader Canadian conversation about AI policy that this guide is contributing to but cannot resolve.

---

What's actually on the table

CHAPTER RECAP — you now have: - The three highest-leverage convergent mechanisms — comprehensive likeness rights legislation, mandatory AI transparency framework, Indigenous data sovereignty integration — with the cross-chapter benefits, institutional pathways, political coalitions, and trade-offs engaged. - The federal legislative mechanisms — AIDA successor, PIPEDA modernization, AI environmental disclosure — with the specific gaps they would address and the political conditions they would require. - The federal-provincial coordination pathway with its historical precedents, what coordination could achieve, the trade-offs in autonomy versus comprehensiveness, and the timeline considerations. - The sectoral regulator empowerment pathway with the specific regulators that could be expanded — OPC, CRTC, OSFI, Health Canada, Competition Bureau, Auditor General, provincial commissioners — and the pathway's strengths and limits. - The international engagement pathway with the specific outstanding commitments — Council of Europe Framework Convention, Beijing Treaty ratification, EU alignment, Sovereign Technology Alliance development. - The judicial pathway with what it can and cannot deliver and the strategic considerations for pursuing legislation versus litigation. - The political-economy conditions honestly engaged — current government's investment-rather-than-regulation choice, opposition party positioning, civil society and advocacy alignment, industry diversity, labour positions, Indigenous engagement — with the structural finding that the barrier is political coordination rather than technical or institutional infeasibility. - The chapter's explicit refusal to commit to specific recommendations beyond the documentation of what could be done — preserving the guide's methodological commitment to making contested questions visible rather than resolving them prematurely.

The next chapter (Chapter 21) returns to the guide's bias-mapping methodology from Chapter 1 and Chapter 19 and shows readers how to navigate the ongoing AI conversation beyond the specific issues this guide has engaged. The conversation will continue; readers carrying the guide's analytical infrastructure will be better-equipped to engage what comes next.

You can now read any Canadian AI policy proposal (government commitment, opposition critique, civil society advocacy, industry position, academic recommendation) with the equipment to recognize which gaps it addresses, which institutional pathway it engages, what political coalition it would require, what trade-offs it involves, and what it does not address. The Canadian AI policy conversation in 2026 contains many voices making many proposals. The proposals exist within the institutional and political landscape this chapter has mapped, and engaging them productively requires understanding the landscape rather than just evaluating the proposals individually.

---

Bias label for this chapter: constructive policy analysis with explicit refusal to commit to specific recommendations. Author lean: methodological commitment to making policy trade-offs visible rather than resolving them; sympathetic to the convergent mechanisms identified across multiple chapters; willing to name political-economy conditions honestly while declining to predict political outcomes; explicit that the manual treats this as constructive analysis rather than as advocacy for any specific political coalition or policy package. Federal and provincial government framings (current policy choices, opposition positioning) treated as primary on the political-economy analysis. Civil society advocacy organizations (CUPE, ACTRA, the Writers' Union of Canada, CIGI, the Office of the Privacy Commissioner, the Indigenous data sovereignty organizations, the BC + AI coalition) treated as primary on specific proposed mechanisms. Academic peer-reviewed analysis (the broader policy literature) treated as primary on theoretical and comparative framing. International frameworks (EU, UK, US, Australia, Singapore, Council of Europe) treated as primary on comparator analysis.

Primary sources cited or relied on in this chapter: AI for All strategy launch documentation (June 4, 2026); Bill C-27 / AIDA documentation (2022–2025); Treasury Board of Canada Secretariat AI Register (November 28, 2025); CUPE Senate Brief (March 2026); Writers' Union of Canada submissions on AI training; ACTRA NIL Rights submission and 2025 Independent Production Agreement; Centre for International Governance Innovation analysis on likeness rights and deepfake legislation; First Nations Information Governance Centre OCAP documentation; Inuit Tapiriit Kanatami National Inuit Strategy on Research; Global Indigenous Data Alliance CARE Principles; Abundant Intelligences program documentation; New Frontiers of Research Fund Transformation grant documentation; European Union AI Act (effective August 1, 2024); European Union Energy Efficiency Directive 2024/1364; Council of Europe Framework Convention on AI (2024); Beijing Treaty on Audiovisual Performances (2012); Quebec Law 25 documentation; Ontario Working for Workers Act (effective January 1, 2026); Pan-Canadian Framework on Clean Growth and Climate Change (precedent for federal-provincial coordination). Detailed citations in the Sources appendix.

---