Artificial Intelligence and the Law

Course Description

Artificial Intelligence and the Law is an introduction to the intersection of artificial intelligence and the law, focusing on its challenges, opportunities, ethics, and impact on the law, the legal industry, and society.

Within the SCNS taxonomy, PLA is the Paralegal prefix. Daytona State publishes this at 3 credits, offered spring, giving approximately 48 contact hours — the ratio established by the published PLA1003 at the same college (3 credits / 48 hours).

This is one of the newest courses in this repository, and it addresses a genuinely live problem rather than a speculative one. Lawyers have been sanctioned for filing briefs containing citations that generative AI invented; firms are deploying AI tools while regulators work out what supervision means; and the technology is simultaneously the most useful and the most dangerous thing to arrive in legal support work in a generation. A paralegal who understands both what these tools do well and how they fail is substantially more employable than one who does not — and substantially less likely to end a career.

Learning Outcomes

Required Outcomes

• Describe artificial intelligence, machine learning, and generative AI at a working level.
• Distinguish rule-based systems, predictive models, and generative models and their different failure modes.
• Describe how large language models produce output and why they fabricate.
• Describe current and emerging applications of AI in legal practice.
• Describe AI-assisted legal research and its verification requirements.
• Describe technology-assisted review and predictive coding in electronic discovery.
• Describe contract analysis, due diligence, and document automation tools.
• Evaluate an AI tool for suitability, reliability, and risk in a legal context.
• Apply professional responsibility rules to the use of AI in legal work.
• Describe the duty of competence as it applies to technology.
• Describe the duty of confidentiality as it applies to AI tools and third-party services.
• Describe supervision obligations for AI-assisted work product.
• Describe the unauthorized practice of law as it relates to AI legal services.
• Describe verification obligations for AI-generated citations and factual claims.
• Describe court rules and standing orders addressing AI use in filings.
• Describe algorithmic bias, its sources, and its consequences in legal contexts.
• Describe AI in criminal justice, including risk assessment and its controversies.
• Describe intellectual property questions raised by AI, including training data and authorship.
• Describe privacy and data protection issues raised by AI systems.
• Describe liability questions arising from AI-caused harm.
• Describe evidentiary issues, including deepfakes and authentication.
• Describe the developing regulatory landscape at state, federal, and international levels.
• Describe the effect of AI on legal employment and on access to justice.
• Use AI tools appropriately and document their use.

Optional Outcomes

• Describe prompt construction and its effect on output quality.
• Describe retrieval-augmented approaches and grounded legal research tools.
• Describe AI governance policies within an organization.
• Describe AI in alternative dispute resolution.
• Describe autonomous systems and product liability.
• Analyze a current case or regulatory development involving AI.

Major Topics

Required Topics

• What AI is: machine learning and generative models
• Model types and their failure modes
• Why language models fabricate
• AI applications in legal practice
• AI-assisted legal research
• Technology-assisted review in e-discovery
• Contract analysis and document automation
• Evaluating tools for legal use
• Professional responsibility and AI
• The duty of technological competence
• Confidentiality and third-party AI services
• Supervision of AI-assisted work
• AI and the unauthorized practice of law
• Verification of AI output
• Court rules and standing orders on AI
• Algorithmic bias
• AI in criminal justice
• Intellectual property: training data and authorship
• Privacy and data protection
• Liability for AI-caused harm
• Evidence, deepfakes, and authentication
• The regulatory landscape
• Employment and access to justice
• Appropriate use and documentation

Optional Topics

• Prompt construction
• Retrieval-augmented and grounded tools
• Organizational AI governance
• AI in dispute resolution
• Autonomous systems and product liability
• Current case and regulatory analysis

Resources & Tools

• The Florida Bar (floridabar.org) — free and the operative authority: the Rules Regulating The Florida Bar, ethics opinions addressing generative AI, and UPL guidance. Florida issued an advisory ethics opinion on lawyers' use of generative AI; read the current version rather than a summary.
• ABA Model Rules and formal ethics opinions — free; the duty of technological competence in the comment to Rule 1.1 is the foundation of this whole area.
• Cornell Legal Information Institute (law.cornell.edu) — free, authoritative, with the Wex dictionary.
• Online Sunshine (leg.state.fl.us) — free full text of the Florida Statutes.
• Florida Courts (flcourts.gov) — court rules and any standing orders; individual judges have issued their own AI disclosure requirements, and checking the specific judge's orders is now part of filing practice.
• Stanford HAI and its legal-hallucination research — free, and the most rigorous public work measuring how often legal AI tools produce wrong answers.
• NIST AI Risk Management Framework — free; the reference framework for evaluating AI risk in an organization.
• EU AI Act materials — free; relevant because it is the most developed regulatory regime and it affects US firms with European clients.
• Westlaw, Lexis, and their AI features — if your programme provides access, use them and compare their output against the underlying authority.
• NALA (nala.org) and NFPA (paralegals.org) — professional associations publishing practical AI guidance for paralegals.
• Court opinions sanctioning AI misuse — free and genuinely the best teaching material in this area; read the actual orders.

Career Pathways

• Paralegal and legal assistant — SOC 23-2011; AI literacy is rapidly becoming an expected competency rather than a differentiator.
• Litigation support and e-discovery specialist — technology-assisted review is now standard in large matters, and this is one of the better-paid paralegal specializations.
• Legal technology specialist — firms and legal departments hire for tool selection, implementation, and training.
• Contract analyst and contract lifecycle management — corporate legal departments.
• Compliance and privacy analyst — data protection and AI governance are growing functions.
• Legal operations — a growing discipline focused on process, technology, and efficiency in legal departments.
• Knowledge management in law firms.
• Legal project management.
• Continuation to a paralegal bachelor's degree or to law school — though note that paralegal credentials are a separate pathway from a J.D. rather than a step toward one.
• The honest employment framing: AI is automating parts of document review and first-draft production that were historically entry-level paralegal work. What is not automating is judgement, verification, client contact, and knowing when the output is wrong — and those are what this course builds.

Special Information

⚠⚠ Rule 11 applies here more than anywhere else in this repository

• This is the fastest-moving area of law covered by any guide in this collection. Court rules, bar ethics opinions, regulatory frameworks, case law, and the tools themselves are all changing on a timescale of months.
• Treat every specific in this guide as of its time, and verify anything operational against a primary source before relying on it.
• Learn the framework, not the current state. The durable content is: what these systems actually do, how they fail, what professional duties apply, and how to verify. Those persist. Specific tools, specific opinions, and specific rules do not.
• Check The Florida Bar's current ethics guidance, the Rules Regulating The Florida Bar, and the specific judge's standing orders for any matter — the last is a genuinely new part of filing practice.
• Your textbook is probably out of date. A course in this area should be reading current opinions and orders, and if it is not, supplement it yourself.
• Be sceptical of confident claims in both directions. Vendors overstate capability; commentators overstate obsolescence. The measured reality is in the research and the sanctions orders.

⚠⚠ Fabricated citations — the failure that has already ended careers

• Generative models produce fluent, plausible, non-existent case citations. They look correct — proper reporter format, plausible party names, plausible holdings — and they do not exist.
• Lawyers have been sanctioned, fined, and referred to disciplinary authorities for filing briefs containing them, in multiple jurisdictions, and the orders are public and widely reported. Read them; they are the best teaching material in this area.
• The excuse "the AI said so" has been rejected consistently. The filer is responsible for the contents of a filing, and Rule 11-type obligations to make reasonable inquiry apply regardless of what produced the draft.
• Verify every citation against the actual authority. Not against a summary, not against another AI, and not against a search result snippet — pull the case, read it, and confirm it says what the draft claims.
• Verify quotations and pin cites too. A real case can be cited for a proposition it does not support, which is harder to catch and equally serious.
• Grounded and retrieval-based legal research tools reduce but do not eliminate the problem. Independent research has found meaningful error rates even in purpose-built legal AI products — "legal-specific" does not mean "verified."
• Paralegals are frequently the verifier. This is where the role's value concentrates, and doing it thoroughly is a genuine professional contribution.
• Document your verification. A record of what you checked protects you and the firm.
• If you find a fabrication, say so immediately — before filing, and clearly.

⚠⚠ Confidentiality: do not put client information into a public tool

• The duty of confidentiality is broad — it covers essentially everything learned in the course of representation, not only privileged communications — and it binds non-lawyer staff.
• Entering client information into a consumer AI service may disclose it to a third party, and depending on the service's terms may allow it to be retained or used for training. This is a live and serious risk, and it has produced disciplinary attention.
• Read the terms of service and the data handling policy of any tool before it touches client matter. Consumer tiers and enterprise tiers frequently have entirely different terms.
• Anonymizing is harder than it looks. Removing names does not remove identifiability when the facts are distinctive, and a matter description can identify a client to anyone in the industry.
• Use only firm-approved tools. Shadow IT — staff using unapproved tools because they are convenient — is the practical way most breaches happen.
• Privilege can be waived by disclosure to a third party. The consequences of an inadvertent waiver are severe and may not be repairable.
• Conflicts and ethical walls apply to tool use too. A shared AI workspace can cross a screen that was carefully constructed.
• Ask before you use anything. "Is this tool approved for client matter?" is the correct question, every time.

⚠⚠ AI, UPL, and access to justice — the hardest question in the course

• Only a member of The Florida Bar may practise law in Florida, and the unauthorized practice of law is enforced by the Bar and the Florida Supreme Court.
• Whether an AI tool that answers legal questions constitutes UPL is genuinely unsettled, and it is being litigated and regulated differently in different jurisdictions.
• The paralegal's own boundary is unchanged. Using AI to draft does not authorize you to advise, select documents for a member of the public, or represent anyone. The tool does not confer authority any more than a form book does.
• Florida has no intermediate "legal document preparer" licence of the kind some states have, which makes the boundary sharper here.
• Supervision is the mechanism that makes non-lawyer work lawful, and it extends to AI-assisted work product — an attorney who signs a filing is responsible for it however it was drafted.
• The access-to-justice argument is real and deserves honest treatment. Most people with legal problems cannot afford a lawyer, and self-help tools reach people the profession does not. The tension between protecting the public from bad legal advice and denying them any legal help at all is the substantive debate, and a good course takes both sides seriously.
• Regulatory sandboxes and alternative business structures in some jurisdictions are experiments in exactly this, with mixed results so far.
• Rule 11 applies — verify current UPL guidance with The Florida Bar directly.

⚠ Bias, criminal justice, and evidence — where the societal stakes concentrate

• Algorithmic bias is not a hypothetical. Models trained on historical data reproduce the patterns in that data, and where the historical patterns reflect discrimination, the model reproduces the discrimination while appearing objective.
• Risk assessment tools in criminal justice — used in bail, sentencing, and parole decisions — have been the subject of sustained analysis and criticism regarding disparate impact and transparency. The debate about how to define fairness mathematically is itself unresolved, and different reasonable definitions are mutually incompatible.
• Opacity is a due process problem. A defendant who cannot examine how a score affecting their liberty was produced has a real objection, and proprietary systems have resisted disclosure.
• Automation bias compounds it. Humans over-trust machine output, particularly under time pressure, so a tool nominally providing "decision support" often decides.
• Deepfakes and synthetic evidence are an emerging authentication problem. Audio, video, and document forgery have become cheap and convincing, and courts are working out how authentication rules apply. The corollary is equally serious: genuine evidence can now be dismissed as fabricated — the so-called liar's dividend.
• Chain of custody and metadata matter more than ever for digital evidence.
• AI in hiring, lending, housing, and insurance raises discrimination questions under existing law, and enforcement agencies have stated that existing anti-discrimination law applies to algorithmic decisions.
• Approach this content analytically rather than ideologically. The useful skill is being able to ask what the system was trained on, what it optimizes, who is affected, and what recourse exists.

⚠ Intellectual property and liability — unsettled, and moving

• Training data and copyright is actively litigated, with substantial cases pending against model developers over the use of copyrighted works. The outcomes will matter and they are not yet known.
• Authorship of AI-generated output is constrained. The U.S. Copyright Office has issued guidance requiring human authorship for registration, and purely machine-generated material has been held unregistrable. Human contribution matters and the line is being drawn case by case.
• Output can infringe. A model producing something substantially similar to a protected work creates exposure regardless of intent, and some vendors now offer indemnification — read what it actually covers.
• Liability for AI-caused harm is unsettled across product liability, negligence, and professional malpractice frames, and the question of who is responsible — developer, deployer, or user — is the central unresolved issue.
• Professional malpractice exposure is clearer. A lawyer who files a defective AI-generated document is liable in the ordinary way, and insurers have begun asking about AI use.
• Contracts increasingly address AI — representations about AI use, restrictions on inputting confidential material, and allocation of AI-related risk are appearing in engagement letters and vendor agreements.
• Regulation is fragmenting. The EU AI Act, various US state laws, and sector-specific rules create a patchwork, and firms with multi-jurisdictional clients must navigate it.
• Rule 11 applies with maximum force — every item in this section may have changed. Verify against primary sources.

How Florida course levels affect transfer

The first digit of an SCNS number denotes the year of offering, not transferability. Courses at the 1000 and 2000 levels transfer transparently between Florida public institutions, and 3000 to 4000 is unproblematic since both are upper division. The boundary that actually matters is 2000 to 3000, where lower-division credit generally cannot satisfy an upper-division requirement.

PLA2872 is 3 credits and approximately 48 contact hours, offered spring. Expect current case reading, ethics analysis, tool evaluation, and written work rather than a fixed textbook — this is a subject where the reading list should be substantially newer than any published text.

As a 2000-level course it transfers on the ordinary lower-division basis between Florida public institutions, though paralegal A.S. degrees are applied and do not carry the A.A.'s junior-status guarantee. The practical value is immediate: firms are actively looking for staff who can use these tools competently and, more importantly, who know precisely when not to trust them.