SUBJECT: The Concept of ‘Artificial Intelligence’ in Legal Research Ethics

I. Introduction

This memorandum provides an exhaustive analysis of the concept of artificial intelligence within the framework of legal research ethics in the Philippines. The integration of AI-powered tools—such as large language model chatbots, predictive analytics platforms, and automated research databases—into legal practice presents novel ethical challenges that intersect with longstanding duties under the Code of Professional Responsibility and Accountability (CPRA). The core ethical tension lies in balancing the efficiency gains of AI with the lawyer’s non-delegable duties of competence, diligence, and confidentiality. This memo will define AI in the legal context, examine applicable ethical rules, analyze specific risks, and propose guidelines for ethically compliant use.

II. Definition and Scope of ‘Artificial Intelligence’ in Legal Research

For the purposes of this ethical analysis, artificial intelligence in legal research refers to computational systems designed to perform tasks traditionally requiring human cognitive functions, such as understanding natural language, recognizing patterns in legal data, predicting outcomes, and generating textual content. Key applications include: (1) AI-assisted legal research platforms that go beyond keyword search to semantic understanding; (2) predictive tools that analyze case law to forecast judicial decisions; (3) document review and e-discovery algorithms; and (4) generative AI that can draft legal memoranda, pleadings, or contract clauses. The ethical analysis applies irrespective of the tool’s sophistication, from simple machine learning models to advanced generative AI.

III. Governing Ethical Framework: The Code of Professional Responsibility and Accountability

The primary source of ethical obligations is the Code of Professional Responsibility and Accountability (Supreme Court Administrative Matter No. 22-09-01-SC). While the CPRA does not explicitly mention artificial intelligence, its core canons and rules provide the governing principles. The lawyer’s duty to society (Canon I), the duty to the courts (Canon II), and the duty to clients (Canon III) are all implicated. Critical specific rules include:
Rule 1.01 on competence* and the duty to keep abreast of legal developments.
Rule 1.02 and 1.03 on diligence* and the requirement of thoroughness in preparation.
Rule 2.01 on candor* and fairness to the court.
Rule 3.05 on confidentiality* of client information.
* Rule 5.07 on the duty to properly supervise subordinates and delegated work.
The ethical use of AI requires interpreting these traditional duties in a new technological context.

IV. Core Ethical Duty of Competence (Rule 1.01, CPRA)

Rule 1.01 mandates that a lawyer shall not undertake a matter without adequate preparation. The use of AI implicates competence in two key ways. First, a lawyer must possess a reasonable understanding of the AI tools they employ—not necessarily the technical programming, but their fundamental functions, limitations, and potential for error (e.g., hallucinations in generative AI, algorithmic bias in predictive tools). Ignorance of these limitations constitutes a failure of competence. Second, competence requires that the lawyer oversees and validates the AI’s output. The lawyer cannot blindly rely on AI-generated research or arguments. The duty of competence ultimately remains personal and non-delegable to the machine.

V. Core Ethical Duty of Diligence and Thoroughness (Rule 1.02 & 1.03, CPRA)

Closely linked to competence is the duty of diligence. Rule 1.03 requires a lawyer to “act with reasonable diligence and promptness in representing a client.” AI can enhance diligence by enabling more comprehensive research. However, it can also undermine it if used as a shortcut that bypasses critical analysis. Ethical diligence demands that the lawyer performs a due diligence review of the AI’s process: verifying the sources it cites, checking the continued validity of cited jurisprudence, and ensuring the research is complete and not skewed by the algorithm’s inherent limitations. The duty of candor to the tribunal (Rule 2.01) is also relevant, as submitting AI-generated content with fictitious citations would violate this duty.

VI. Core Ethical Duty of Confidentiality (Rule 3.05, CPRA)

Rule 3.05 strictly protects all confidential information relating to a client’s representation. Inputting client data, case details, or strategy into a third-party AI platform poses a significant confidentiality risk. Many AI systems use user inputs to further train their models, potentially making confidential information part of the system’s database and accessible to other users. A lawyer must obtain the client’s informed consent after consultation before using an AI tool that requires disclosing confidential information, unless the tool is proven to have robust, contractually guaranteed data isolation and privacy protocols. The duty to protect confidentiality extends to the lawyer’s choice of technology vendor.

VII. Duty of Supervision and Accountability

Rule 5.07 of the CPRA states that a lawyer “shall be responsible for the work product of his or her associates, apprentices, and non-lawyer staff.” By analogy, this principle of supervision and ultimate accountability applies to the use of AI tools. The AI system is a tool used by the lawyer or under the lawyer’s direction. The lawyer cannot attribute an error to the “machine.” The ethical obligation is to establish and maintain appropriate supervisory procedures over the use of AI, including training for staff, implementing verification protocols, and ensuring final, human lawyer review and approval of all work product. The following table compares traditional research ethics with AI-augmented research ethics:

Ethical Dimension	Traditional Legal Research	AI-Augmented Legal Research
Duty of Competence	Mastery of manual research in reporters, digests, and known databases.	Understanding AI tool functions, limitations, and appropriate use cases.
Duty of Diligence	Physically checking cited sources for accuracy and context.	Actively verifying AI-generated citations and conducting independent checks for algorithmic bias or gaps.
Confidentiality	Securing physical files and using privileged communication channels.	Scrutinizing AI vendor data processing agreements, understanding data retention policies, and avoiding input of confidential information.
Supervision	Direct oversight of non-lawyer staff and junior associates.	Implementing technical and procedural safeguards for AI use and mandating human review checkpoints.
Work Product	Clearly human-generated, with known sources.	Risk of undetected hallucinations or plagiarism if not rigorously audited.
Accountability	Lies squarely with the lawyer for all work product.	Lies squarely with the lawyer; the AI is not a scapegoat for ethical lapses.

VIII. Specific Ethical Risks: Hallucinations, Bias, and Unauthorized Practice of Law

Beyond the core duties, specific risks arise. First, hallucinations: generative AI may invent plausible-sounding but non-existent case law, statutes, or quotes. Relying on such output violates duties of competence, diligence, and candor to the tribunal. Second, algorithmic bias: if training data is skewed, AI may produce research that systematically overlooks certain perspectives or jurisprudence, leading to inadequate representation. Third, the unauthorized practice of law: while AI is a tool, if a lawyer allows an AI to make substantive legal judgments or direct legal strategy without meaningful lawyer intervention, it may blur the line of proper delegation. The lawyer must remain the principal actor.

IX. Recommended Best Practices for Ethical Compliance

To mitigate risks, lawyers should adopt the following best practices:

Due Diligence on AI Tools: Vet the AI vendor’s reputation, security, and training data policies.

Training: Educate oneself and all firm personnel on the capabilities and limitations of specific AI tools.

Informed Consent: Consider obtaining client informed consent for the use of AI, especially where confidential information is involved.

Human-in-the-Loop Protocol: Mandate that a qualified lawyer must review, verify, and take responsibility for all AI-generated output. Use AI for augmentation, not automation.

Input Sanitization: Avoid inputting sensitive confidential information or personally identifiable information. Use hypotheticals or anonymized data where possible.

Continuous Monitoring: Stay updated on jurisprudence and ethics opinions regarding AI from the Supreme Court and the Integrated Bar of the Philippines.

X. Conclusion

The concept of artificial intelligence in legal research does not create new ethical duties but radically transforms the environment in which existing duties under the Code of Professional Responsibility and Accountability must be fulfilled. The ethical lawyer must approach AI not as an autonomous agent but as a sophisticated tool requiring rigorous oversight. The duties of competence, diligence, confidentiality, and supervision demand a proactive, educated, and cautious integration of AI into practice. Failure to adapt these core ethical principles to the AI context risks professional liability, malpractice, and disciplinary action. Ultimately, the ethical burden of ensuring accurate, confidential, and diligent legal research remains irrevocably with the lawyer.