Growth & Pro — automated fact-checking

Catch what AI engines
invent about your brand.

AI hallucination detection for brand facts — pricing, features, founders, HQ, founding year. Every scan fires five targeted fact-check prompts per engine and classifies each finding by severity.

5

Fact types checked per engine per scan: pricing, founders, HQ, features, founding year

5-tier

Severity classification — from mostly right to defamatory

Daily

Automated scanning cadence on Growth and Pro plans

Automated detection

Five prompts. Five fact types.
Every scan, every engine.

01

Record your ground truth

Enter your verified brand facts: current pricing tiers, founding year, headquarters city, top three features, and founder names. These become the benchmark every AI response is measured against.

02

Five prompts fire per engine

Every scan sends five targeted fact-check queries to each AI engine on your plan — one per fact type. Each prompt is crafted to elicit the engine's actual belief about that specific claim.

03

Claude Haiku classifies each finding

Every response is evaluated against your ground truth and assigned a severity level: mostly right, outdated, confused, fabricated, or defamatory. You receive a timestamped report for every discrepancy.

Evidence you can act on

Timestamped findings.
Direct quotes.

Every finding includes the exact text each AI engine produced, the fact type it contradicts, and your recorded ground truth side-by-side. Forward this directly to the AI provider's correction channel or share with your PR team.

  • Engine name, fact type, and timestamp for every finding
  • Exact AI-generated quote — screenshot-ready evidence
  • Your verified ground truth displayed alongside for comparison
  • Severity classification from mostly right to defamatory
  • Email notification when new hallucinations are detected
Hallucination Report — 2026-06-10
[FABRICATED]·ChatGPT·Pricing

AI said: "Plans start at $9/month with a free tier for individual users. Business pricing from $29/month for up to 10 seats."

Your fact: No free tier. Starter plan begins at $49/month.

[OUTDATED]·Perplexity·Headquarters

AI said: "Headquartered in Austin, Texas, with additional engineering offices in New York."

Your fact: Relocated to San Francisco, CA in Q3 2024. Austin office closed.

[MOSTLY RIGHT]·Gemini·Founding Year

AI said: "Founded in 2018 by a team of former enterprise software engineers."

Your fact: Founded in 2019. Founder backgrounds differ from the AI description.

Illustrative example — your actual report reflects your brand facts

Five-tier severity

Not all hallucinations are equal.

Each finding is classified so your team knows exactly how urgently to act — and what specific action to take.

Mostly Right

What it means

Minor inaccuracy embedded in an otherwise correct description. The AI has the broad picture right but one specific detail wrong.

Recommended action

Monitor for drift. Clarify the specific detail in your public content and documentation.

Outdated

What it means

Information was accurate at a prior point in time but now reflects a previous state of your business.

Recommended action

Update your content with current information, add a visible revision date, and submit to key third-party directories.

Confused

What it means

Your brand's facts are mixed up with a competitor or another company operating in your space.

Recommended action

Strengthen differentiation messaging. Publish explicit comparison content and ensure your entity data is unambiguous.

Fabricated

What it means

The AI produced specific details — numbers, names, claims — with no factual basis in any known source.

Recommended action

File a correction through the AI provider's feedback mechanism. Document all instances with timestamps.

Defamatory

What it means

False statements that could meaningfully damage your brand's reputation or business relationships.

Recommended action

Escalate to legal counsel immediately and preserve all evidence. File corrections with every affected engine.

Five fact types

The claims AI gets wrong most often.

Pricing

Wrong tiers, invented free plans, fabricated monthly or annual costs

Features

Capabilities you don't have, integrations that don't exist, workflow claims

Founders

Invented names or backgrounds attributed to your founding team

Headquarters

Wrong city, country, or office location details

Founding Year

Incorrect incorporation date, launch year, or company age

Real consequences

Wrong AI facts damage real pipelines.

Pricing errors

most common hallucination type

AI engines frequently invent pricing tiers, free plans, and cost figures that don't match your actual offering — arriving in sales conversations as buyer expectations your team has to undo.

Feature claims

second most common type

AI engines describe integrations and capabilities your product doesn't support, creating support tickets, failed trials, and churn from expectations your product never set.

Undetected

without systematic scanning

AI model knowledge updates without warning. A correct description today may become a fabrication after the next fine-tune. You'll only discover it when a prospect quotes it back at you.

Scenarios described are illustrative examples based on the hallucination categories GEOscanAI detects.

Plan availability

Included on Growth and Pro.

Starter
Growth
Pro

Hallucination scanning included

Scanning + email alerts

Start monitoring

Know what AI engines say about you.

14-day free trial. No credit card required. Hallucination detection on Growth and Pro.

Start your free trial →