Before you send that AI-generated demand letter, catch the errors that could destroy your credibility and weaken your legal position
7Categories of AI Errors
100+Letters Reviewed
14+Years Experience
$47KLargest Failed AI Letter
⚠️ Does Your AI Letter Need Professional Review?
Quick Risk Assessment
Answer these questions to determine if your ChatGPT demand letter is safe to send:
1. What's the amount in dispute?
Under $1,000
$1,000 - $5,000
$5,000 - $10,000
Over $10,000
2. How complex is your contract?
Simple consumer terms/refund policy (1-2 pages)
Standard service agreement (3-5 pages)
Complex commercial contract (5+ pages)
Partnership/IP/Employment agreement
3. Did ChatGPT cite any specific laws or statutes?
No legal citations
Yes, cited specific statutes or regulations
Yes, cited case law or court decisions
Yes, multiple legal citations throughout
4. Who are you sending this to?
Individual person or small business
Mid-size company
Large corporation or well-known brand
Payment processor (Stripe/PayPal) or platform
5. Have you verified every factual claim in the letter?
No, I trusted ChatGPT's draft
I spot-checked a few things
Yes, verified every date, amount, and fact
The $47,000 Mistake
A tech founder sent a ChatGPT-drafted demand letter to Stripe citing "Federal Banking Regulation 482.3(b)" – a statute that doesn't exist. Stripe's response: "Your cited statute does not exist. We are closing this matter." Once you send a demand letter with your name on it, you can't unsend it.
🚨 The 7 Categories of AI Errors
Click each category to see detailed examples and how to fix them:
📅
Factual Errors & Assumptions
AI fills gaps with invented details that sound plausible but aren't true.
❌ BAD: AI Embellishment
"Despite numerous attempts over several weeks to reach customer service..."
✓ GOOD: Verified Facts
"I contacted customer service on October 15, 2024 via email (attached) and received no response."
Fix: Remove any "numerous," "repeatedly," "multiple times" unless you can document each instance with dates and evidence.
⚖️
Hallucinated Legal Citations
Most dangerous: AI invents statutes, regulations, and case law that don't exist.
Fix: Verify EVERY citation on official government websites. If you can't verify it in 5-10 minutes, remove it. It's safer to cite your contract than fake law.
📚
Misapplied Legal Theories
Real laws cited incorrectly or in wrong context.
Invoking consumer protection laws for B2B disputes
Claiming "breach of fiduciary duty" with no fiduciary relationship
Demanding punitive damages for simple breach of contract
Threatening "criminal fraud charges" for civil disputes
Fix: Stick to breach of contract claims based on your actual contract language. Don't let AI add creative legal theories you can't support.
📋
Ignored Contractual Terms
AI doesn't check for arbitration clauses, damage caps, or mandatory procedures.
Critical Error Pattern
AI threatens "immediate court action" when contract requires binding arbitration. This signals you didn't read the contract and don't have legal counsel.
Fix: Search your contract for "arbitration," "AAA," "JAMS," "dispute resolution," "limitation of liability," "notice," and "cure period." Make your letter consistent with these terms.
💰
Unrealistic Remedies
Demanding damages, fees, or relief that aren't available.
"Treble damages" (only available under specific statutes)
"All attorney's fees" (rarely available without contract clause)
"Punitive damages" (not available for contract breach)
Fix: Focus on actual damages you can prove. Only demand fees if contract or statute specifically provides for them.
😤
Tone Problems
Either overly formal legalese or unnecessarily aggressive posturing.
❌ TOO AGGRESSIVE
"Your willful, malicious, and egregious violations constitute systematic fraud. We will ensure your business reputation is destroyed."
✓ PROFESSIONAL & FIRM
"You failed to deliver the services specified in our contract, causing documented damages of $X. I request payment by [date] or I will pursue arbitration as required under Section 12 of our agreement."
🔍
Missing Critical Components
AI forgets basic structural elements that make a demand letter effective.
Clear, specific demand (what exactly do you want?)
Realistic deadline with specific date
Articulation of next steps if ignored
Reference to supporting documents
Proper header with contact information
🔍 Three-Pass Audit Process
Audit your AI-generated letter in three focused passes, each targeting different types of errors:
Pass 1: Factual Accuracy
Verify party names: Pull your contract and check exact legal entity names
Audit timeline: Create a factual timeline from your documents. Compare every date in the letter against your records
Check all numbers: Every dollar amount must match documentation. Create a spreadsheet showing your calculation
Verify quotes: If letter quotes someone, make sure they actually said that (check emails/recordings)
Pro Tip: Reverse-Engineer Your Facts
Before editing the AI letter, write out your own bullet-point fact summary on paper: just dates, amounts, who said what, and when. Force the AI letter to match YOUR fact summary, not the other way around.
Pass 2: Legal Accuracy
Highlight every legal citation: Mark every statute, regulation, case, or "federal law requires" reference
Verify each citation: Google "[citation] official text" and confirm it exists and says what AI claims
Check contract first: Your strongest arguments come from the contract, not statutes. Quote actual contract language
Arbitration reality check: Search contract for "arbitration." If found, letter should reference arbitration, not court
Audit remedies: Attorney's fees, treble damages, punitive damages – only demand what's actually available
Pass 3: Strategic Review
Verify all essential components: Header, background, breach, harm, demand, deadline, next steps, supporting docs
Assess tone: Read aloud. Imagine it in court. Remove ALL-CAPS, multiple !!!, personal attacks
Match realistic next step: Are you actually prepared to file arbitration/small claims if ignored?
Consider relationship: One-time dispute vs. ongoing business relationship affects tone
📖 How to Verify Legal Citations
ChatGPT Cannot Access Legal Databases
ChatGPT doesn't look up laws – it predicts what a legal citation should look like based on patterns. This results in confident hallucinations of statutes that sound real but don't exist.
Step-by-Step Citation Verification
STEP 1: Extract All Citations
Use a highlighter (or different color in your word processor) and mark:
• Specific statutes: "California Civil Code § 1542"
• Federal regulations: "15 U.S.C. § 1692"
• Case names: "Smith v. Jones (2023)"
• General references: "federal consumer protection law requires"
STEP 2: Google Each Citation
For each citation, search: "[exact citation] official text"
Look for results from:
• Official government websites (.gov)
• State legislature sites
• Cornell Legal Information Institute (law.cornell.edu)
• Justia or FindLaw (established legal databases)
STEP 3: Verify It Says What AI Claims
Read the actual statute. Does it:
• Actually exist? (Some AI citations are complete fiction)
• Say what the letter claims? (AI sometimes cites real laws incorrectly)
• Apply to your situation? (Consumer law cited for B2B dispute?)
• Is it current? (Not repealed or substantially amended?)
The 5-Minute Rule
If you cannot verify a citation within 5-10 minutes of reasonable searching, remove it. It's safer to make your argument based on contract language and facts than to risk citing fake law.
Common AI Citation Hallucinations
AI Citation
Problem
What to Do
"California Civil Code § 2847.5"
Section doesn't exist
Remove entirely or find actual relevant statute
"Federal Banking Regulation 482.3(b)"
Complete fabrication
Remove. No such regulation exists.
"FTC requires 30-day refunds"
Oversimplified/wrong context
Check actual FTC regulations for your situation
"Under federal law, you must..."
Vague, unsupported
Specify actual law or remove claim
✅ Final Pre-Send Checklist
Before sending your demand letter, verify every item below. Click each item as you complete it:
Checklist Progress
0/17 Complete
Factual Verification: Every fact has been verified against documentation
Party Names: All party names and addresses are correct and match contract
Amounts: All dollar amounts are accurate and calculation is explained
Timeline: Timeline is accurate and supported by dated evidence
Citations: Every legal citation has been verified or removed
Contract Terms: Contract terms are quoted accurately (not paraphrased)
Remedies: Demanded remedies are actually available under law/contract
Arbitration: Letter acknowledges arbitration if contract requires it
Tone: Tone is professional and strategic (no personal attacks/threats)
Threats: All threats are realistic ones you're prepared to follow through on
Specific Demand: What you want is crystal clear and specific
Deadline: Realistic deadline with specific date is included
Next Steps: What happens if demand is ignored is clearly stated
Supporting Docs: Supporting documents are referenced and attached
Format: Letter is on appropriate letterhead or professional format
Proofreading: Spelling and grammar are correct
Documentation: You've kept a copy and documented when/how you sent it
✓ Checklist Complete!
You've verified all critical items. Your letter is significantly less risky than most AI-generated demand letters. However, if your dispute involves high stakes, complex contracts, or any of the red flags in the next tab, professional review is still recommended.
🚩 Red Flags: When You MUST Get Professional Review
Some situations are too risky to rely on self-audited AI letters, even if you follow every step. Seek professional review if ANY of these apply:
💰
High Financial Stakes
Dispute over $10,000 or any amount that would materially impact your finances or business.
Copyright, trademark, patent, trade secret claims all have specific legal requirements. Don't DIY this.
🏢
Large Companies/Brands
Public companies and well-known brands have experienced legal teams. Amateur letters signal lack of representation.
🌎
Multiple Jurisdictions
Parties in different states/countries, or demands against multiple related entities require professional analysis.
⚔️
Counterclaim Risk
If there's risk the recipient might sue you in response, you need defensive strategy before sending demands.
Professional Letter Review & Rewrite
Send me your ChatGPT draft and supporting documents. I'll provide a corrected, legally sound version that preserves your narrative while fixing common AI errors.
It depends on stakes and complexity. For very small, straightforward consumer disputes ($300 refund with clear return policy), a carefully audited ChatGPT letter may be reasonable. But larger amounts, complex contracts, employment issues, IP matters, or anything that could result in litigation requires professional review. If you wouldn't feel comfortable representing yourself in small claims court, don't send a DIY demand letter.
This damages your position significantly. When a lawyer sees hallucinated statutes or impossible remedies, it tells them: (1) you don't have legal representation, (2) you don't understand your case, and (3) they can likely ignore your letter. This is why the audit process is critical. Remove any citations you cannot personally verify. Better to have a simple, fact-based letter than one full of fake law that gets immediately dismissed.
Absolutely not. There's no strategic benefit and significant downsides. It signals lack of legal representation and may cause them to scrutinize for AI errors. ChatGPT is a drafting tool like Microsoft Word – you wouldn't announce "I wrote this in Word." The letter should stand on its substantive merits. However, IF you're working with a lawyer, absolutely tell them you used AI and provide your prompts.
Yes, potentially. Attorney-client privilege protects communications with your lawyer, not with third-party AI services. When you input information into ChatGPT, you're sharing with OpenAI subject to their terms. Don't put privileged communications from your attorney into AI tools. Don't share sensitive business information or strategy. If working with a lawyer, let them handle AI interaction or work under their guidance on what can be safely input.
Don't panic. If they point out technical errors but don't address your substantive claim, acknowledge the correction professionally and refocus on the merits. If their response indicates serious legal problems, consult a lawyer before responding further. Don't try to dig yourself out with another AI-generated response. Sometimes recipients' lawyers are overly aggressive about minor errors hoping you'll go away – a lawyer can assess whether their response is substantive or tactical posturing.
All current AI models share the same fundamental limitations: they predict plausible text but don't understand law, cannot verify facts, and cannot exercise strategic judgment. Claude tends to be more conservative and less prone to invented citations, but still hallucinates legal content. The audit process in this guide applies regardless of which AI you used. The tool matters less than the verification process.
Honest self-assessment: If you can't clearly explain your legal theory and what law/contract provision supports your demand, it's too complex. Other indicators: contract over 5 pages, multiple parties involved, disputed facts, prior hostile communications, significant amount relative to your resources, time pressure/deadlines, or you're hoping to establish precedent for other cases. The test: If someone asked you to walk them through your legal theory, could you do it clearly and confidently? If not, get professional help.