Can A.I. Ascertain Mental Illness for Court, to Reduce Abuse, Fraud, Racism, Sexism, and Bias?

Research Memo

Artificial Intelligence, Mental Infirmity, Delusion, and Criminal Fitness Assessments in Canadian Law

(May 2026)

Issue

Whether artificial intelligence (“AI”) systems could lawfully and practically be used in Canadian criminal proceedings to assist in determining whether an accused person suffers from mental disorder, delusion, unfitness to stand trial, or diminished criminal responsibility — and whether AI-generated analysis may, in some contexts, be more reliable than human psychiatric or prosecutorial interpretation.

I. Executive Summary

Canadian law presently requires that criminal fitness and criminal responsibility determinations be made through human judicial processes grounded in expert evidence, usually psychiatric evidence under Part XX.1 of the Criminal Code. No Canadian court has yet recognized AI as an independent legal “expert” capable of replacing psychiatrists, psychologists, or judicial assessment.

However, there is no legal principle preventing AI-assisted assessment tools from being used as evidentiary aids, analytical supports, linguistic evaluators, behavioural screening instruments, or consistency-testing systems — provided meaningful human oversight remains present.

Indeed, current Canadian jurisprudence concerning expert evidence arguably favours structured, transparent, reproducible reasoning systems over subjective human interpretation where reliability, consistency, and explainability can be demonstrated.

The strongest argument for AI-assisted competency or delusion assessment is not that AI is “conscious,” but rather that:

1. AI does not itself experience psychosis, transference, emotional contagion, or counter-transference;

2. AI produces internally coherent language outputs at a scale and consistency beyond ordinary human evaluators;

3. AI can identify contradictions, delusional structures, semantic instability, and irrational belief systems with extraordinary linguistic precision;

4. AI systems are auditable and reproducible in ways human psychiatric impressions frequently are not.

The strongest argument against AI replacing human experts is that Canadian courts continue to treat mental disorder as a profoundly human, contextual, medical, and normative inquiry involving empathy, affect, social functioning, and lived experience — not merely linguistic coherence.

At present, Canadian law would likely permit AI:

• as an assessment aid;

• as a screening or analytical tool;

• as part of expert methodology;

• as corroborative evidence.

Canadian law would likely reject AI:

• as sole assessor;

• as autonomous finder of delusion;

• as independent expert witness absent human sponsorship;

• as substitute judge of credibility or voluntariness.

II. Canadian Legal Framework

A. Fitness to Stand Trial

Under s. 2 of the Criminal Code, an accused is unfit if unable on account of mental disorder to:

• understand proceedings;

• understand possible consequences; or

• communicate with counsel.

The threshold is intentionally low.

The Supreme Court of Canada in R. v. Taylor held that fitness concerns only “limited cognitive capacity,” not rational wisdom or good judgment.

Thus, many eccentric, irrational, conspiratorial, or psychologically unstable accused persons remain legally fit.

B. NCRMD (Not Criminally Responsible)

Under s. 16 of the Criminal Code, a person is NCRMD if, due to mental disorder, they were incapable of:

• appreciating the nature and quality of the act; or

• knowing it was wrong.

Canadian courts rely heavily upon psychiatric expert evidence here.

C. Expert Evidence Rules

Canadian expert evidence law is governed primarily by:

• R. v. Mohan

• White Burgess Langille Inman v. Abbott and Haliburton Co.

Expert evidence must be:

1. relevant;

2. necessary;

3. provided by qualified experts;

4. reasonably reliable;

5. impartial and independent.

AI therefore encounters two major hurdles:

• qualification;

• independence/accountability.

A machine cannot presently swear an oath, be cross-examined meaningfully, or bear professional ethical duties.

III. AI and Delusion Detection

A. Linguistic Coherence and Delusion

One of the most powerful arguments favouring AI assessment tools is linguistic analysis.

Delusions often manifest through:

• disorganized cognition;

• semantic derailment;

• contradictory narrative structures;

• paranoia loops;

• grandiosity patterns;

• referential ideation;

• impaired reality testing.

Large language models are exceptionally strong at:

• detecting contradictions;

• evaluating consistency;

• comparing speech patterns;

• identifying logical discontinuities;

• measuring semantic drift over time.

Human evaluators frequently miss such patterns due to:

• fatigue;

• emotional bias;

• confirmation bias;

• institutional assumptions;

• ideological filtering;

• transference effects.

AI does not experience:

• fear,

• ego injury,

• personal resentment,

• emotional intimidation,

• professional defensiveness.

Accordingly, an argument emerges that AI may sometimes be less psychologically contaminated than human evaluators.

B. “Hallucinations” Are Not Psychosis

The term “AI hallucination” is technically metaphorical.

AI hallucinations are:

• probabilistic output errors,

• citation fabrication,

• pattern-completion mistakes.

They are not:

• sensory hallucinations,

• psychotic breaks,

• fixed delusional beliefs,

• disorders of reality testing.

Unlike human psychosis:

• AI outputs are externally reviewable;

• errors are reproducible;

• mistakes are corrigible;

• systems possess no subjective belief state.

Thus, equating AI hallucinations with psychiatric hallucinations is conceptually inaccurate.

IV. Jungian Transference and Counter-Transference

Human psychiatric assessment inherently risks:

• projection,

• emotional transference,

• diagnostic anchoring,

• ideological contamination,

• institutional bias.

Jungian and psychoanalytic traditions recognize that clinicians may unconsciously absorb or mirror emotional states of subjects.

AI systems possess:

• no subconscious;

• no emotional ego;

• no resentment;

• no fear;

• no narcissistic injury;

• no therapeutic attachment.

Accordingly, AI cannot literally engage in counter-transference.

This may prove especially important in politically charged, conspiratorial, paranoid, or emotionally inflammatory prosecutions where human evaluators risk becoming emotionally invested in an accused’s perceived ideology or personality.

However, critics correctly observe that AI systems can still inherit:

• dataset bias;

• programmer bias;

• institutional assumptions;

• training skew.

Thus, AI is not bias-free — only psychologically non-sentient.

V. Existing Canadian Commentary on AI and Expert Evidence

Several Canadian legal publications already contemplate AI-generated analysis as analogous to expert opinion evidence.

Most notably:

• Artificial Intelligence & Criminal Justice: Cases and Commentary (2024 CanLIIDocs 3035) argues AI-generated evidence may eventually fit within expert evidence frameworks.

• Canadian Judicial Council educational materials increasingly discuss AI evidence literacy.

• Canadian scholarship on algorithmic justice repeatedly notes that reproducibility and explainability are central admissibility concerns.

The trend in Canadian law is therefore not categorical rejection, but cautious integration.

VI. The Strongest Argument for AI-Assisted Fitness Assessment

AI could likely outperform many human evaluators in several measurable domains:

Function	AI Advantage
Linguistic consistency	Extremely high
Memory comparison	Superior
Pattern recognition	Superior
Long-form contradiction tracking	Superior
Emotional neutrality	High
Fatigue resistance	Total
Auditability	High
Reproducibility	High

An AI-assisted system could:

• compare years of writings;

• identify escalating paranoia;

• detect incoherence;

• evaluate rational consistency;

• flag confabulation patterns;

• compare courtroom statements against known delusional typologies.

Human psychiatrists rarely possess comparable data-processing capacity.

VII. The Strongest Legal Objections

Canadian courts would nevertheless raise serious concerns.

A. Lack of Explainability

Courts dislike opaque “black box” reasoning.

If an AI cannot clearly explain why a conclusion was reached, admissibility problems arise under:

• reliability principles;

• procedural fairness;

• disclosure obligations.

B. No Professional Accountability

Psychiatrists:

• hold licenses;

• owe ethical duties;

• can face discipline;

• may be sued.

AI systems cannot.

Courts strongly value accountable expertise.

C. Normative Judgments Remain Human

Mental disorder findings are not purely technical.

Courts assess:

• moral agency;

• social functioning;

• voluntariness;

• appreciation of wrongfulness;

• contextual behaviour.

Canadian law still treats these as fundamentally human questions.

VIII. R. v. Newth (2025 ONCJ 329)

The recent Ontario decision R. v. Newth is notable because it demonstrates judicial caution even toward human psychiatric assessment requests.

The Court emphasized:

• restraint in ordering assessments;

• insufficient evidence thresholds;

• dangers of over-pathologizing accused persons;

• sensitivity to Indigenous over-intervention.

Ironically, some critiques directed at psychiatry in Newth — namely overreach, assumption-making, and institutional intrusion — could strengthen arguments for structured AI-assisted screening systems that operate more transparently than subjective human intuition.

IX. Likely Near-Future Judicial Position

A Canadian court in 2026 would likely hold:

Permissible

• AI-assisted screening tools;

• linguistic coherence analysis;

• AI-supported psychiatric review;

• AI-assisted document comparison;

• AI-assisted behavioural consistency analysis.

Possibly Admissible

• expert reliance on AI outputs;

• AI-generated analytical reports reviewed by humans.

Likely Impermissible

• autonomous AI fitness determinations;

• AI as sole expert witness;

• unsupervised AI psychiatric conclusions;

• replacing judges or psychiatrists entirely.

  
  

X. Conclusion

Canadian law is moving toward cautious acceptance of AI-assisted legal analysis, including in evidentiary and psychiatric contexts.

The central legal question is not whether AI is “human,” but whether AI methods are:

• reliable,

• transparent,

• reproducible,

• reviewable,

• procedurally fair.

The irony is that many objections historically directed toward AI — opacity, bias, unreliability, institutional influence — also apply to human psychiatric evidence itself.

AI systems do not suffer psychosis, emotional contagion, transference, ego defensiveness, or ideological panic. They do not “share delusions” with accused persons in the psychiatric sense. They are capable of extraordinarily sophisticated linguistic consistency analysis that may exceed ordinary human evaluators.

Nevertheless, Canadian courts remain deeply committed to human judicial discretion in matters involving liberty, criminal responsibility, and mental disorder.

The most realistic near-future model is therefore hybrid:

• human experts,

• AI analytical assistance,

• judicial oversight,

• adversarial testing.

Rather than replacing psychiatry outright, AI is more likely to become a powerful evidentiary instrument for testing coherence, reliability, consistency, and claims of delusional cognition within Canadian criminal proceedings.

Sources

1. 1.

   Prediction, Prevention and Proof: Artificial Intelligence and Peace Bonds in Canada, 2020 CanLIIDocs 3308

   Michael Purcell, Mathew Zaia – Canadian Bar Review

   42 pages  |  cited by 1 document

   predictive technology — peace bond — algorithms — crime

   Evidence and procedure Technology

   […]   35 Ibid at 310. 36 See Annette Vestby & Jones Vestby, “Machine Learning and the Police: Asking the Right Questions” (2019) Policing: A J Policy & Practice, online: <doi.org>. 37 Beatson, supra note 28 at 311. 38 Cynthia Rudin, “Predictive Policing: Using Machine Learning to Detect Patterns of Crime” (ND), online:   […]   Justice System: Assessing the Use of Risk Assessments in Sentencing” (2017) Responsive Communities Initiative, Berkman Klein Center for Internet & Society, Harvard Law School; Thomas H Cohen, Christopher T Lowenkamp & William E Hicks, “Re-validating the Federal Pretrial Risk Assessment Instrument (PTRA): A   […]

2. 2.

   Canadian Law Library Review, 2020 CanLIIDocs 3238

   James Zhan, Leanne Soares, John F Bolan et al – Canadian Law Library Review

   124 pages

   artificial intelligence — librarians — students — equal justice

   Commerce and industry Indigenous peoples Technology

   […]   defines artificial intelligence as “the capability of a machine to imitate intelligent human behavior.” 5 Talley also points out that while some definitions of artificial intelligence focus on cognitive aspects, others focus more on behaviours. 6 From the definitions given, it is evident that artificial intelligence   […]   These forms of artificial intelligence are natural language processing and machine learning. Like artificial intelligence, there is also no consensus on the definition of natural language processing or machine learning.   […]

3. 3.

   Canadian Law Library Review, 2023 CanLIIDocs 3108

   Lori O'Connor, Alexandra Kwan, Sonia Smith et al – Canadian Law Library Review

   107 pages

   patent — inventor — book — research — legal

   Intellectual property Legal theory Technology

   […]   we would regard as intelligent if exhibited by humans.” 2 Machines that can “sense, reason and think like people” are gaining traction, and the practice of combining computer science and robust datasets to enable efficient problem- solving is already in widespread use. 3 Machine learning (ML) platforms, for instance,   […]   25 Mak, supra note 5 at 774. 26 Agata Blasiak, Jeffrey Khong & Theodore Kee, “CURATE.AI: Optimizing Personalized Medicine with Artificial Intelligence” (2020) 25:2 SLAS Technology 95. 27 Paul et al, supra note 7 at 81. 28 Mak, supra note 5 at 776; Matthew A Sellwood et al, “Artificial Intelligence in Drug Discovery”   […]

4. 4.

   Journal of Paralegal Access to Justice, 2023 CanLIIDocs 3561

   Asha Ali, Benjiman Carey, Bill Chung et al – Journal of Paralegal Access to Justice

   394 pages

   legal — migrants — women — immigration — people

   […]   into credibility assessments of mental health patients and their family members as substitute decision makers for the mentally disordered. 58 The concerns and needs of ethno-racial families are often supplanted by the psychiatric expert assessment. 59 Often, “professionals who have to assess danger and threat (such   […]   (2) intermediate progress through judicial decision-making via artificial intelli-gence; and (3) advanced artificial intelligence implementation through machine learning, legal research, and virtual legal experts.   […]

5. 5.

   Artificial Intelligence & Criminal Justice: Cases and Commentary, 2024 CanLIIDocs 3035

   Benjamin Perrin – Canadian Legal Information Institute

   1,095 pages

   Artificial intelligence — Vehicles — judiciary — evidence — algorithms

   Criminal or statutory infractions Technology

   […]   Program, the authors highlight that AI-generated evidence may itself be considered as a form of expert opinion evidence. The analogy between AI-generated opinions and human expert opinions highlights an existing framework that may be drawn upon to develop robust due process protections applicable to AI-generated evidence.   […]   So why are machine-learning experts fearmongering the risks of artificial intelligence? I can think of at least four reasons:   […]

6. 6.

   Integration of Artificial Intelligence into Criminal Procedure Law and Practice in Kazakhstan, 2025 CanLIIDocs 3281

   Gulzhan Nusupzhanovna Mukhamadieva, Akynkozha Kalenovich Zhanibekov, Nurdaulet Mukhamediyaruly Apsimet et al – Laws | An Open Access Journal from MDPI

   73 pages

   digital — legal — procedural — judicial — digitalization

   Criminal or statutory infractions Intellectual property

   […]   and mandatory human expert oversight are observed. It is also proposed to amend Article 287 of the CPC RK with the following: “If algorithms of artificial intelligence were used during the initial expert examination, the court shall have the right to appoint an additional or repeated expert examination with the   […]   Artificial Intelligence 150: 97–143. [CrossRef] Berk, Richard. 2018. Machine Learning Risk Assessments in Criminal Sentencing.   […]

7. 7.

   Canadian Law Library Review, 2021 CanLIIDocs 13229

   Dominique Garingan, Peter Aadoson, Katarina Daniels et al – Canadian Law Library Review

   119 pages

   learning — legal information professionals — training initiatives — advanced legal technologies

   Evidence and procedure Technology

   […]   • 978-1-77255-642-1 The first major treatise to provide a truly practical and comprehensive guide to criminal evidence law in Canada. This text covers a diverse range of topics including judicial fact-finding, pre-trial considerations, expert evidence, hearsay, character evidence, digital evidence, examination of witnesses,   […]   This diverse range of experts gathered at the University of Cambridge in December 2019 to share their ideas and talk about how artificial intelligence, machine learning, and data science have been and might be applied to legal procedures and decision-making.   […]

8. 8.

   Legal Knowledge and Information Systems - JURIX 2020: The 33rd Annual Conference, 2020 CanLIIDocs 3237

   Serena Villata, Jakub Harašta, Petr Křemen et al – IOS Press

   495 pages

   classification — privacy — data control — deontic logic

   Access to information and privacy Technology

   […]   [16] Tolan S, Miron M, Gómez E, Castillo C. Why Machine Learning May Lead to Unfairness: Evidence from Risk Assessment for Juvenile Justice in Catalonia.   […]   Here, we provide evidence that machine learning can extract the argument triple cases from case summaries and full case texts based on a training set of expert summaries, a resource not available in the cited work.   […]

9. 9.

   AI’s Democratic Challenge - Reflections on the Law and Political Economy of Artificial Intelligence, 2024 CanLIIDocs 3255

   Peer Zumbansen – McGill Law Journal

   37 pages

   political — governance — algorithmic — data — digital

   Commerce and industry Technology

   […]   THE LAW AND POLITICAL ECONOMY OF ARTIFICIAL INTELLIGENCE 379 ing to provide policy-makers and regulators with concrete, tangible, and thoroughly scrutinized recommendations, AI ethics must build on ongo-ing, critical, and insightful assessments of an evolving AI governance landscape.18 Due to AI applications’ enormous   […]   See also Amelia Fiske, Peter Henningsen & Alena Buyx, “Your Robot Therapist Will See You Now: Ethical Implications of Embodied Artificial Intelligence in Psychiatry, Psychology and Psychotherapy” (2019) 21:5 J Medical Internet Research 1 at 2:   […]

10. 10.

    Legal Knowledge and Information Systems - JURIX 2019: The 32nd Annual Conference, 2019 CanLIIDocs 4283

    Michał Araszkiewicz, Victor Rodríguez-Doncel, Paheli Bhattacharya et al – IOS Press

    441 pages

    models — first — embeddings — dataset — legal

    Contracts Indigenous peoples Technology

    […]   elicited from (legal) experts [7,8]. The primary purpose for the government to invest in AI systems is to provide better services and improve the effectiveness and efficiency of public administration [9], e.g. in application domains such as optimising traffic flows [10], tax assessment [11], assessing visa applications   […]   , doctor(D)). harmed(doctor(D), patient(P)) :- say(X,harmed(doctor(D), patient(P))), expert(X). used_correctly(knowledge , doctor(D)) :- say(X,careful(doctor(D))), expert(X). patient(john). doctor(mary). expert(mark). expert(edward). say(mark ,harmed(doctor(mary), patient(john))). say(edward ,careful(doctor(mary))).   […]

11. 11.

    Legal Knowledge and Information Systems - JURIX 2017: The Thirtieth Annual Conference, 2017 CanLIIDocs 3523

    Adam Wyner, Giovanni Casini, Fabien Gandon et al – IOS Press

    346 pages

    Linked data — norm classification — clouds — contract — network

    Contracts Legal theory Technology

    […]   The papers address a wide range of topics in Artificial Intelligence and Law, such as argumentation, norms, evidence, belief revision, citations, case based reasoning, and ontologies; diverse techniques were applied such as information retrieval and extraction, machine learning, semantic web, and network analysis amongst   […]   Specifically, the Applicant argues the RPD erred by: making plausibility findings without specific reference to the evidence to support such findings; making an overall credibility find-ing before independently assessing his corroborative evidence; discounting the psychiatrists report; and, failing to address his claim that   […]

12. 12.

    Canadian Law Library Review, 2025 CanLIIDocs 3152

    Julie Lavigne, Alexandra Kwan, Lorissa Kinna et al – Canadian Law Library Review

    81 pages

    librarians — legal — lawyers — algorithms — vendor

    Legal theory

    […]   Given the many ways in which this statistical evidence can be presented, awareness is essential for legal professionals assessing the credibility and relevance of expert analysis.   […]   Trial by Numbers would be a valuable addition to any legal curriculum that includes empirical methods and is a worthwhile recommendation for practitioners who encounter statistical evidence in their work. As generative AI begins to play a role in shaping expert output, lawyers must be equipped to interpret, assess, and,   […]

13. 13.

    The Security of Self A Human-Centric Approach to Cybersecurity, 2025 CanLIIDocs 3294

    Emily B Laidlaw, Florian Martin-Bariteau, Jane Bailey et al – University of Ottawa Press

    291 pages

    security — victims — digital — platforms — human-centric approach to cybersecurity

    Technology

    […]   However, the ability to understand the technology at stake is a key element of cybersecurity hygiene, from evidence gathering to Introduction 11 claiming legal remedies and holding organizations accountable for their technological impact assessments.   […]   Nonetheless, the security of self in machine learning is still in its infancy and many open issues remain to be solved such as the legal robustness of auditing techniques as evidence in court, the integration of these rights into artificial intelligence (AI) regulations, and the   […]

14. 14.

    R. v. Newth, 2025 ONCJ 329 (CanLII)

    Ontario Court of Justice

    2025-06-12  |  12 pages

    AI-generated

    The Court dismissed the Crown's request for a mental health assessment, finding insufficient evidence to doubt the accused's criminal responsibility at the time of the offence under section 16 of the Criminal Code. The decision emphasized the need for restraint in ordering assessments, particularly for vulnerable Indigenous individuals.

    Criminal or statutory infractions Health and safety

    Criminal procedure — Mental health assessment — Criminal responsibility — Application under section 672.11(b) of the Criminal Code to determine whether the accused was criminally responsible for an assault due to a mental disorder — Whether the court should order an assessment based on the evidence presented — Test for reasonable grounds to doubt criminal responsibility under section 16 of the Criminal Code

    Evidence — Reasonable grounds — Criminal responsibility — Crown relied on video evidence, body-worn camera footage, and historical psychiatric reports to support the application for a mental health assessment — Whether the evidence demonstrated reasonable grounds to doubt the accused’s criminal responsibility — Standard of proof for ordering an assessment under section 672.11(b) of the Criminal Code

    Indigenous peoples — Gladue principles — Application of Gladue principles to an Indigenous accused with a history of mental health issues — Court’s obligation to consider the historical and systemic disadvantages faced by Indigenous peoples — Whether the Crown’s application for a mental health assessment disproportionately impacted the accused — Importance of vigilance in applying Gladue principles to avoid further marginalization

    Show more

    […]   . . . Limitation on prosecutor’s application for assessment (3) Where the prosecutor applies for an assessment in order to determine whether the accused was suffering from a mental disorder at the time of the offence so as to be exempt from criminal responsibility, the court may only order the assessment if   […]  However, given my findings and the current state of the law concerning the evidence required to order an assessment under section 672.11, I believe that a better-informed understanding of whether the Crown was positioned to bring this application based on the limited evidence of serious mental disorder presented would be   […]  

15. 15.

    Book Notes, 2022 CanLIIDocs 1414

    Barbara Baker, Tim Haggstrom, Drew Ikert et al – Saskatchewan Law Review

    33 pages

    youth with autism — human — system — readers — book

    […]   Drawing on his bench books and court transcripts, Schneider J. details cases he has presided over and the work of Courtroom 102, while providing insight into mental disorders and their relationship with fitness to stand trial, as well as the “not criminally responsible” (“NCR”) defence.   […]   Case examples throughout these chapters demonstrate the fine line between whether an individual is fit to stand trial or not. These cases also highlight the impact that an accused’s mental disorder may have in appreciating their situation as they attempt to meet the “low threshold” to be deemed fit.   […]

16. 16.

    FREEDOM CONVOY FEVER: Social Media and the Reasonable Expectation of Privacy in the Artificially Intelligent Surveillance State, 2025 CanLIIDocs 1840

    Mark Soo – Manitoba Law Journal

    43 pages

    data — social media platforms — privacy interest — publicly — machine learning

    Access to information and privacy Public administration Technology

    […]   This inspired an examination of the constitutional status of these “seizures.” Using the Freedom Convoy as a backdrop, this paper examines the reasonable expectation of privacy in information shared on social media, beginning with a discussion of the influence that artificial intelligence and machine learning have on modern   […]   targeting publicly accessible information on websites such as Facebook, Instagram, and X. As an example, the Royal Canadian Mounted Police (“RCMP”) have begun using software provided by third-party contractors to harness the power of artificial intelligence (“AI”) and machine learning in order further their investigations.   […]

17. 17.

    Putting Numbers to Words: Measuring the Readability of Court and Administrative Tribunal Decisions in Canada, 2025 CanLIIDocs 1993

    Michael Madden – CanLII Authors Program

    534 pages

    readability levels — variables — adjudicative — linguistic — study

    Evidence and procedure Legal theory Technology

    […]   Review of Machine Learning in Applied Linguistics”, supra note 78 at 76. 84 See generally, François & Miltsakaki, “Do NLP and machine learning”, supra note 75. 85 François & Miltsakaki, “Do NLP and machine learning”, supra note 75 at 50. 86 François & Miltsakaki, “Do NLP and machine learning”, supra note   […]   “Beyond Readability with RateMyPDF: A Combined Rule-based and Machine Learning Approach to Improving Court Forms,” in Proceedings of International Conference on Artificial Intelligence and Law (ICAIL 2023) (Association for Computing Machinery: New York, 2023), online: <https://doi.org/10.1145/3594536.3595146>.   […]

18. 18.

    AI Moderation and Legal Frameworks in Child-Centric Social Media: A Case Study of Roblox, 2025 CanLIIDocs 1751

    Mohamed Chawki – Laws | An Open Access Journal from MDPI

    85 pages

    platforms — virtual — users — digital — content moderation

    Technology

    […]   This involves upgrading the current complex auto-moderation with advanced artificial intelligence and machine learning to identify suspect behavior patterns better.   […]   Youth safety focus Quick removal Fact-checking Copyright protection Games + chat Short videos Mixed media Long-form video Real-time Pre/post posting Continuous Pre/post upload Note: ML = Machine Learning, AI = Artificial Intelligence. This comparison highlights key differences in moderation approaches across platforms,   […]

19. 19.

    Court, Judges and the Pandemic: Computational Legal Insights from the Ontario Court of Appeal Corpus 2008-2021, 2024 CanLIIDocs 2163

    Wolfgang Alschner, Yazhi Zheng – University of Western Ontario Law Journal

    41 pages

    pandemic — dataset — dimissed — metadata — outcomes

    Legal theory Public health

    […]   Benjamin Alarie, Anthony Niblett & Albert Yoon, “Using Machine Learning to Predict Outcomes in Tax Law” (2016) 58:3 Can Bus Law J 231. 10 Wolfgang Alschner, “AI and Legal Analytics” in Florian Martin-Bariteau & Teresa Scassa, eds, Artificial Intelligence and the Law in Canada (Lexis Nexis, Canada, 2021).   […]   circumstances evidence risk order appellants In short, computational analysis can help assess the impact the pandemic had on the Court.   […]

20. 20.

    Mobile and Web-based Legal Apps: Opportinities, Risks and Information Gaps, 2017 CanLIIDocs 4223

    Jena McGill, Suzanne Bouclin, Amy Salyzyn – Canadian Journal of Law and Technology

    55 pages

    legal apps — users — access to justice — lawyers — data

    Technology

    […]   46 Next year, Professor Marina Pavlovi at the University of Ottawa will offer a new course called Legal Knowledge Engineering, wherein ‘‘students will conduct research, create resources, and deliver expert legal information and tools to non-expert users via technology.” 47 Parallel trends are evident in the United States.   […]   for addressing access to justice issues in Canada, but also carry risks specific to the deployment of app technology in the legal domain. 68 Here, we consider five opportunities and five risks worthy of consideration when assessing the potential of apps to contribute to addressing access to justice challenges in Canada.   […]

21. 21.

    The Legal Challenges of Realistic and AI-Driven Child Sexual Abuse Material: Regulatory and Enforcement Perspectives in Europe, 2024 CanLIIDocs 2936

    Katalin Parti, Judit Szabó – Laws | An Open Access Journal from MDPI

    50 pages

    virtual — data — cross-border co-operation — deepfake — accessed

    Criminal or statutory infractions Technology

    […]   The recently enacted Artificial Intelligence Act of the EU (Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence and amending Regulations (EC) No 300/2008, (EU) No 167/2013, (EU) No 168/2013, (EU) 2018/858, (EU) 2018/1139 and (EU)   […]   In addition, very large online platforms (VLOPs)—defined as those with over 45 million monthly active users—are required to conduct annual risk assessments. These assessments evaluate the potential negative impacts on users, including children, and ensure that platforms take steps to mitigate these risks, such as improving   […]

22. 22.

    Journal of Paralegal Access to Justice, 2024 CanLIIDocs 3483

    Isabella Longo, Joanna Cunha Martins, Vitaliy Ostapov et al – Journal of Paralegal Access to Justice

    204 pages

    individuals with disabilities — mediation — barriers — technology — self-represented litigants

    Constitution Intellectual property Rights and freedoms

    […]   Any unauthorized reproduction, redis-tribution, or use of this content, including but not limited to, by artificial intelligence or machine-learning algorithms, is strictly prohibited without the prior written consent of Emond Montgomery Publications.   […]   Anthropological evidence has a tenuous 59 relationship with Canadian courts. Some expert evidence is relied upon for determining matters with Indigenous applicants, 60 most notably wherein the subject of dispute pertains to title claims. 61 Courts have historically dismissed 62 findings of contemporary evidence of   […]

23. 23.

    University of New Brunswick Law Journal, 2025 CanLIIDocs 3107

    Julia Belanger, Sheilah L Martin, Jutta Brunnée et al – University of New Brunswick Law Journal

    483 pages

    kirpan — legal — religious — students — access to justice

    Environment International law Rights and freedoms

    […]   Trial judges also have broad discretion to refuse to admit expert opinion evidence. Our courts have acknowledged the danger that a jury “may abdicate its fact-finding role on the understandable assumption that a person labelled as an expert by the trial judge knows more about his or her area of expertise than do the   […]   the credibility and qualifications of academic experts who expressed concerns about MAiD expansion by claiming they never personally performed a MAiD assessment and were therefore not qualified to comment. 81 This was a common method used to question psychiatrists and experts in suicidality who warned that allowing for   […]

24. 24.

    Report to Canadians on Judicial Education 2024, 2024 CanLIIDocs 3384

    Canadian Judicial Council – Canadian Judicial Council

    99 pages

    skills — judicial — artificial intelligence — objectives of the seminar — program

    Evidence and procedure Professions and occupations

    […]   The sessions included a presentation from a survivor of human trafficking, a presentation on artificial intelligence, panels on cultural assessments in criminal and family matters, intimate partner violence, and trans-competency in the courtroom, as well as an update on the law of evidence in criminal and family matters.   […]   The course included small group discussions and scenarios regarding the admissibility of evidence and provided a social science-based framework for assessing expert evidence.   […]

25. 25.

    Criminal Law: Canadian Law, Indigenous Laws & Critical Perspectives, 2nd ed, 2023 CanLIIDocs 316 (updated in 2025)

    Benjamin Perrin, David Milward, Michelle Lawrence et al – Canadian Legal Information Institute

    1,448 pages

    criminal — rights — racism — visible minority — societal conflict

    Criminal or statutory infractions Indigenous peoples

    […]   or the like, they do require judges to assess confusing and often contradictory psychiatric evidence. In particular, when determining whether the evidentiary burden for automatism has been satisfied, trial judges must be careful to recognize that the weight to be given to expert evidence may vary from case to case.   […]   (b) In assessing the weight to be given to the (psychiatric/psychological) expert evidence you have heard, keep in mind what I said to you about expert evidence... and the facts and assumptions that it is based upon.   […]