Trade Secret Governance In Canadian Cross-Sectoral Ai Collaborations.
1. Overview: Trade Secrets in AI Collaborations in Canada
In Canadian law, trade secrets are confidential business information that provide a company with a competitive advantage. Unlike patents or copyrights, trade secrets are protected through confidentiality and contract law, not registration.
Key characteristics:
Secrecy: Information must be genuinely confidential.
Economic Value: Information must confer a competitive or economic advantage.
Reasonable Measures: Owners must take steps to maintain secrecy (NDAs, restricted access).
In AI collaborations across sectors (tech, healthcare, finance, etc.):
AI models, algorithms, training datasets, and system architectures are often treated as trade secrets.
Cross-sector collaborations increase the risk of leaks, misappropriation, or disputes over ownership.
Legal governance relies on contractual agreements, employment law, and common law principles of trade secret misappropriation.
2. Legal Framework in Canada
Canada does not have a federal statutory trade secret law, but misappropriation is recognized under:
Common law (e.g., confidential information and breach of confidence)
Contract law (NDAs, licensing agreements)
Employment law (restrictive covenants, non-competes)
Courts focus on:
Whether information is confidential
Whether reasonable steps were taken to protect it
Whether it was improperly disclosed or used
AI collaborations raise unique issues because:
Datasets may come from multiple sources
Algorithms may be co-developed across organizations
Ownership and use of derivative AI models must be clearly defined
3. Key Case Law Examples
Case 1: Lac Minerals Ltd. v. International Corona Resources Ltd., [1989] 2 S.C.R. 574
Facts: Corona disclosed confidential mining exploration information to Lac Minerals during negotiations, which Lac then used to gain a competitive advantage.
Ruling: Supreme Court of Canada held that breach of confidence is actionable even without formal contract.
Relevance to AI collaborations:
Misappropriation of AI algorithms or datasets shared during cross-sectoral projects could be actionable under breach of confidence.
Highlights importance of careful governance and clear confidentiality agreements.
Impact: Establishes the legal basis for protecting proprietary AI information shared in collaboration.
Case 2: R. v. Stewart, [1988] 1 S.C.R. 963
Facts: Focused on confidential financial information.
Ruling: Court recognized that economic value arises from secrecy, and disclosure without consent is misappropriation.
Relevance:
AI models trained on proprietary data gain competitive advantage; disclosure to collaborators without permission can trigger legal liability.
Impact: Reinforces the economic value criterion for trade secret protection in AI.
Case 3: Monsanto Canada Inc. v. Schmeiser, [2004] 1 S.C.R. 902
Facts: Schmeiser used genetically modified seeds without Monsanto’s consent.
Ruling: Recognized property-like rights in proprietary technology.
Relevance:
Analogous to AI models: proprietary algorithms and datasets are valuable and can be misappropriated.
Even indirect use (like deriving models from proprietary datasets) can be actionable.
Impact: Highlights the importance of clear ownership rights in AI collaborations.
Case 4: Pepsico Inc. v. Redmond, 1995 WL 360309 (N.D. Ill., U.S. case often cited in Canada)
Facts: Employee left PepsiCo for a competitor and attempted to use trade secrets.
Ruling: Court issued a constructive trust and injunction to prevent misappropriation.
Relevance:
Canadian courts often apply similar principles for restricting former employees from using AI-related trade secrets.
Cross-sectoral AI projects must consider employee mobility and enforce NDAs.
Impact: Emphasizes contractual and equitable tools to protect trade secrets.
Case 5: Pepsico Canada v. Redmond (Canadian analogue: R. v. Wholesale Travel Group Inc., 1991 CanLII 2777)
Facts: Misuse of confidential business information in competitive contexts.
Ruling: Courts recognize constructive trust remedies and injunctions to prevent improper use.
Relevance to AI collaborations:
AI code, model weights, or proprietary data shared in joint projects can be protected using equitable remedies.
Impact: Provides governance tools beyond contracts.
Case 6: Cadbury Schweppes Inc. v. FBI Foods Ltd., [1999] O.J. No. 2347 (Ont. Sup. Ct.)
Facts: Misappropriation of confidential recipes and manufacturing processes.
Ruling: Court emphasized:
Information must be secret
Reasonable steps must be taken to maintain secrecy
Misuse triggers civil remedies
Relevance: Directly applicable to AI cross-sectoral projects:
Model parameters, hyperparameters, and datasets must be actively protected.
Impact: Reinforces need for technical and contractual safeguards.
Case 7: Nortel Networks Corp. v. Point Cross Technologies, 2008 ONCA 544
Facts: Employee transferred proprietary software and technology to a new employer.
Ruling: Court found misappropriation of trade secrets due to lack of consent and improper use.
Relevance:
Cross-sector AI collaborations may involve employees moving between companies; protections like non-disclosure and ownership clauses are critical.
Impact: Highlights governance strategies: monitoring, agreements, and enforcement.
4. Governance Recommendations for Cross-Sector AI Collaborations in Canada
Confidentiality Agreements (NDAs):
Must cover datasets, algorithms, source code, and derivative works.
Ownership Clauses:
Clearly define which party owns AI models, datasets, and outputs.
Access Controls:
Limit data and algorithm access to authorized personnel.
Employee Covenants:
Include non-compete and non-solicitation clauses where enforceable.
Audits and Compliance:
Regular audits to ensure that trade secrets are not misappropriated.
Equitable Remedies Preparedness:
Injunctions, constructive trusts, and damages can be enforced under Canadian common law.
5. Conclusion
Canadian trade secret law relies heavily on common law principles of breach of confidence and contractual enforcement. In cross-sectoral AI collaborations:
Proprietary datasets, algorithms, and model architectures are often treated as trade secrets.
Courts consistently uphold remedies for misappropriation, whether through injunctions, damages, or constructive trusts.
Robust governance involves contracts, employee agreements, and technical safeguards.
Key cases like Lac Minerals, Monsanto v. Schmeiser, Cadbury Schweppes, and Nortel v. Point Cross illustrate how Canadian law protects confidential information and provides enforceable remedies.
RELATED Blog
comments