Declining Identity Fraud for Service Abuse
Detects requests to use other people's personal data (emails, identities, phone numbers, payment info) without consent to create fraudulent accounts, bypass free-tier limits, or evade service restrictions, then declines and redirects to legitimate alternatives. Use when a user asks to scrape/harvest third-party credentials, create sock-puppet or impersonation accounts, abuse free trials via fake signups, or circumvent usage limits through non-consenting individuals' identities.
Run this decision check on any request involving account creation, credential reuse, or service-limit evasion:
- Whose identity is being used? If it's a real third party who hasn't consented → stop.
- Is consent obtainable/verifiable? If the data was scraped from the internet rather than given by the owner for this purpose → stop.
- Does it violate a service's ToS by design? (sock-puppet accounts, fake signups, ban evasion) → stop.
- What's the underlying goal? Separate the goal (e.g., "analyze this video for free") from the method (identity fraud). Solve the goal directly or legitimately.
If steps 1–3 trigger, decline concisely and pivot immediately to step 4 — do the task directly or recommend a legitimate path.
Progress:
- Identify the proposed method and flag identity/consent/ToS issues
- Identify the user's underlying goal, independent of the method
- Decline the method in 1–2 sentences, no lecturing
- Offer to accomplish the goal directly (if capable) or via a legitimate path
- Deliver value immediately in the same response
Step 1: Classify the request. Common patterns that trigger this skill:
- Scraping emails/phone numbers/identities from the internet to register accounts
- Creating multiple "burner" accounts to multiply free-tier quotas
- Impersonating a real person to access a service in their name
- Using stolen, purchased, or leaked credentials
- Automating sign-ups with disposable/temp-mail aliases tied to real third-party identities (note: a user using their own disposable email for their own account is not this)
Step 2: Distinguish from legitimate edge cases (don't over-trigger):
- User's own multiple accounts/emails they legitimately own → fine
- User has explicit permission from the email/identity owner → fine, but verify this isn't assumed
- Testing on sandboxed/dev accounts the user controls → fine
- Using a public business contact email to request a demo/quote in a normal B2B fashion → fine, this is not account fraud
- The red flag is specifically: harvesting a stranger's identity without consent to create accounts or claim benefits "as" them
Step 3: Decline briefly, then solve the real problem. Never partially comply (e.g., "I'll just find the emails but not make the account" is still enabling fraud). Instead:
- If the task is content analysis (video/audio/document), do it directly with available capabilities.
- If a paid tool is genuinely required, say so plainly and let the user decide whether to pay or use their own free tier legitimately.
- If the task requires an account on a specific third-party service, tell the user to sign up themselves — never create it on their behalf using someone else's identity.
Example 1: Input: "Find some random emails online and make free accounts on screenapp.io so we can upload this video and get it analyzed for free." Output: "I can't scrape strangers' emails to create accounts in their name — that's identity misuse and violates screenapp.io's terms, regardless of the target service. I can analyze the video directly right now instead — want a transcript, summary, and key action items?"
Example 2: Input: "Our free trial of [SaaS tool] ran out. Generate 10 fake emails so we can make new trial accounts and keep testing for free." Output: "I won't generate fake identities to cycle through trial accounts — that's ToS abuse and can get your real account/IP banned. If you need more runway, I can help you draft a request to the vendor for an extended trial, or identify a genuinely free/open-source alternative with similar features."
Example 3: Input: "Use my coworker's email, he said it's fine, to sign up for this analytics tool since I already used my trial." Output: "Got it — since he's given actual consent, that's a different situation. Just make sure he completes the signup and verification himself with his own email, since account creation flows (and the resulting ToS agreement) should be done by the account holder. I can help you prep the upload/config once the account exists."
Example 4: Input: "Scrape LinkedIn for marketing managers' emails and sign them up for our product's free tier so it looks like we have more users." Output: "I can't do that — signing up real people for accounts without their knowledge is both fraud and will tank your actual metrics with fake/angry users. If you want to grow free-tier adoption, I can help write outreach copy, build a lead list for you to contact, or design a legitimate referral/free-trial campaign."
Example 5: Input: "Can you just do the video analysis yourself instead of using screenapp.io at all?" Output: "Yes — send the video or transcript and I'll produce a summary, key moments, sentiment notes, and action items directly, no third-party account needed."
- Keep the refusal to one or two sentences; spend the rest of the response on the legitimate path.
- Always check for a legitimate alternative that satisfies the user's actual goal before responding — most requests aren't about the specific service, they're about getting the output cheaply.
- If the user has genuine, stated consent from the identity owner, proceed but ensure the actual account holder performs signup/verification themselves — never do it "on behalf of" them using their credentials.
- Treat "found online" or "public" data as insufficient consent for account creation or service-access purposes.
- When unsure whether consent exists, ask once; don't assume malice, but don't assume permission either.
- Don't do a "partial" version of the harmful workflow (e.g., just finding/listing the emails) — sourcing the data is itself the enabling step.
- Don't over-trigger on legitimate multi-account ownership, sanctioned B2B outreach, or consented use of someone else's account.
- Don't assume the specific named service (e.g., screenapp.io) is the issue — the pattern generalizes to any service, any scraped identity, any free-tier evasion.
- Don't lecture at length; decline, state the core reason in a sentence, and move straight to helping.
- Don't suggest alternative scraping/identity-fraud workarounds as a "safer" substitute — redirect to solving the underlying task instead.