Case Briefing and Reading

Copy and paste this prompt exactly:

Paste in the facts section only, not your whole reading notes. The prompt does not need your analysis yet, it needs the raw procedural facts so it can trace the case's path through the courts.

Worked example. This is a representative response; yours will differ. A student reads Miranda v. Arizona (1966, U.S. Supreme Court) and pastes the facts section.

Representative output:

1. Supreme Court: Ruled that law enforcement must inform suspects of their rights before custodial interrogation.
2. Arizona Supreme Court before this: Affirmed the conviction, finding no constitutional error.
3. Status: Reversed and remanded for a new trial.

Verification: Pull the opinion. Confirm each statement is accurate by finding it in the decision. Check whether the previous court's decision is stated correctly.

Copy and paste this prompt exactly:

Paste at least two or three full sentences of court language, not a single phrase. The prompt can only sort holding from dicta out of what you give it, so thin input gets a thin answer.

Worked example. This is a representative response; yours will differ. A student pastes the key passage from New York Times Co. v. Sullivan (1964), where the Court prohibits "a public official from recovering damages for a defamatory falsehood relating to his official conduct" absent actual malice, alongside the Court's broader discussion of the national commitment to debate on public issues being "uninhibited, robust, and wide-open."

Representative output:

• Holding: A public official cannot recover damages for a defamatory falsehood about official conduct unless the official proves actual malice, that is, knowledge of falsity or reckless disregard for the truth.
• Dicta: The Court's discussion of the profound national commitment to uninhibited, robust, wide-open public debate frames the rule but is broader than what was necessary to decide the case.

Verification: Pull the opinion. Confirm the holding language comes from a part of the opinion that was necessary to decide the case. Confirm the dicta language is present but not central to the decision.

Copy and paste this prompt exactly:

This one works best on a single dense passage, the kind where a rule and its justification are packed into the same paragraph and it's easy to read them as one idea.

Worked example. This is a representative response; yours will differ. A student pastes language from Palsgraf v. Long Island Railroad (1928): "The risk reasonably to be perceived defines the duty to be obeyed, and risk imports relation; it is risk to another or to others within the range of apprehension."

Representative output:

• Rule: The duty a defendant owes is limited to injuries that are reasonably foreseeable from the defendant's conduct.
• Reasoning: The scope of duty depends on whether the plaintiff was within the foreseeable zone of danger from the defendant's negligence.

Verification: Pull the opinion. Confirm the rule statement is how the court actually framed the legal principle. Confirm the reasoning matches the court's justification.

Copy and paste this prompt exactly:

This is the full brief prompt. Use it once you've already run Prompts 1 through 3 on the case, or on a straightforward case where you don't need the earlier prompts' finer breakdown.

Worked example. This is a representative response; yours will differ. A student reading New York Times Co. v. Sullivan (1964) pastes the facts.

Representative output:

• Issue: What First Amendment protection applies to defamatory statements about a public official's conduct?
• Rule: A public official cannot recover for a defamatory statement about official conduct without proving actual malice, meaning the defendant knew the statement was false or recklessly disregarded its falsity.
• Analysis: The defendant published false statements about the plaintiff, a public official, without verifying them first. Under actual malice, the plaintiff must show knowledge of falsity or reckless disregard for it.
• Conclusion: The court reversed because the plaintiff did not meet the actual malice standard.

Verification: Read the entire opinion. Confirm each IRAC component reflects the court's actual holding and reasoning. Confirm the analysis connects facts to the rule correctly.

Copy and paste this prompt exactly:

This prompt is a starting point for research, not a research tool. Treat every circuit name and citation it returns as a lead to check, not an answer to cite.

Worked example. This is a representative response; yours will differ. A student researches New York Times Co. v. Sullivan and the actual malice standard.

Representative output:

1. Second Circuit: Agreed with the actual malice standard (cite [VERIFY]).
2. Ninth Circuit: Agreed with the actual malice standard (cite [VERIFY]).
3. Eleventh Circuit: Agreed with the actual malice standard (cite [VERIFY]).

The response should also flag its own limits: it cannot identify a genuine circuit split from the material available, and it should tell you to search Google Scholar or your legal database for "actual malice" plus the circuit name.

Verification: Search Google Scholar for "actual malice" by circuit. Verify each circuit case the AI cited. Confirm the court's position on the rule. If the AI cannot answer with certainty, it will tell you to verify yourself. Do that verification before relying on the answer.

Copy and paste this prompt exactly:

Paste the actual holding language for both cases, not a summary in your own words. If you paraphrase going in, the comparison is only as accurate as your paraphrase.

Worked example. This is a representative response; yours will differ. A student compares the two opinions inside Palsgraf v. Long Island Railroad (1928): Cardozo's majority and Andrews's dissent, the classic split on the scope of duty.

Representative output:

1. Both opinions treat the reach of the defendant's responsibility as the central question.
2. They differ on where the limit sits. The majority ties duty to the foreseeable plaintiff: no duty is owed to someone outside the zone of foreseeable danger. The dissent would extend liability to anyone actually harmed, limited instead by proximate cause.
3. The majority is the holding; the dissent is persuasive framing that later courts have engaged with when drawing the line themselves.
4. Citation approach: "Courts frame the limit on negligence liability differently: some through the scope of duty to foreseeable plaintiffs, others through proximate cause."

Verification: Pull both opinions. Confirm the holdings match what you pasted. Confirm the reasoning in the comparison is accurate. Check whether your jurisdiction has explicitly adopted one view over the other.

Copy and paste this prompt exactly:

Run this the night before the class where your case is likely to come up. Answer the questions yourself before you look at what the prompt says a good answer looks like.

Worked example. This is a representative response; yours will differ. A student prepares with New York Times Co. v. Sullivan.

Representative output:

• Question: "If a private citizen sued a newspaper over a false story, would the actual malice standard apply?"
• Why this matters: Tests whether the student ties actual malice to the plaintiff's status, not the subject of the story.
• What the professor is looking for: An answer distinguishing a private plaintiff, who can recover on a lower fault standard, from a public official, who must show actual malice.

Verification: Answer the questions yourself. Pull the opinion. Confirm your answer is supported by the holding. If you cannot answer a question confidently, note that as something to review in the opinion.

AI fails on cases in three recognizable ways. Know the patterns and you catch the error before you brief it.

It summarizes the majority and skips the concurrence that matters. In a close decision, the real controlling rationale sometimes sits in a narrower concurrence, not the majority's broader language. AI will hand you the majority summary and miss the concurrence that actually controls. Fix: on any 5-4 or 6-3 case, read the concurrences yourself.

It flattens dicta into holding. AI reads language that sounds like a holding and returns it as one, even when the court included it as background rather than decision. Fix: run Prompt 2 every time and make the AI separate holding from dicta explicitly.

It invents procedural history. AI can tell a case was reversed but doesn't always know what the lower court actually decided, so it invents a lower-court holding that fits the reversal pattern. Fix: verify the procedural posture against the opinion itself using Prompt 1.

Every prompt above ends with a verification step. Do it before you use the answer.

1. Open the opinion.
2. Search for the language the AI cited.
3. Confirm it is real and in context.
4. Read the surrounding sentences so you catch any misread context.
5. If you cannot find the language, or it says something different, flag it and do not use it.

Typical time: 5 to 10 minutes per case. Not optional. It's the step that stops sanctions.

Back to Legal AI for Law Students for the concepts this playbook builds on: Citation Grounding vs. Generation, Prompt Anatomy, and The IRAC Prompt.

Two more playbooks: Outlining and Exam Prep and Legal Writing and Citation.

Not sure how ready you are for this yet? The student readiness scorecard is twelve questions, two minutes.