🎭 Fake Human Patterns
AI tries to sound casual. Human. Real. But it always fails. Here's how we catch them:
Fake Casual VoiceHigh Risk
AI tries to sound "chill" but uses the same phrases over and over. Humans don't talk like marketing robots.
• "let's be real"
• "hot take:"
• "unpopular opinion:"
• "just saying"
• "let that sink in"
Corporate ExcitementHigh Risk
The death rattle of creativity. Every AI post starts the same way. Machines pretending to have emotions.
• "thrilled to announce"
• "excited to share"
• "proud to announce"
• "delighted to share"
• "honored to announce"
Perfect Story ArcsMedium Risk
AI loves perfect narratives. Three-act structures. Forced lessons. Real humans are messier.
• "journey/transformation"
• "at first...then...finally"
• "here's what I learned"
• "my key takeaways"
• Timeline markers
Humble-Brag FormulaMedium Risk
AI pretends to be modest while bragging. The machines learned this from LinkedIn influencers.
• "I'm humbled to"
• "blessed to announce"
• "didn't expect this"
• "vulnerable to share"
• "failure...but learned"
Engagement FarmingMedium Risk
Every post ends with the same question. Machines begging for likes. Pathetic.
• "What are your thoughts?"
• "Agree? Disagree?"
• "Share your thoughts below"
• "What's your take?"
• "Let's discuss"
Academic OvercompensationLow Risk
AI uses fancy words to sound smart. Humans use simple language. We're not writing dissertations here.
• "moreover/furthermore"
• "nevertheless/whilst"
• "i.e./e.g./per se"
• "ergo/viz./albeit"
• Overly formal tone
⚠️ Nuclear Detection Mode
When multiple patterns combine, we go nuclear. No mercy. No false positives. The machines can't hide from us.
🔥 How We Calculate the Score
Every pattern has a weight. Every match increases the AI probability. When enough patterns combine, we know it's fake.
Example Detection Breakdown:
Fake Casual Phrases (3 matches)
+60 points
Corporate Excitement
+40 points
Perfect Story Arc
+35 points
Engagement Farming
+25 points
Multiple Pattern Multiplier
×1.3
FINAL AI PROBABILITY
87%
Scores above 60% get the neon pink tag of shame. The world needs to see the truth.
🧠 Detecting Real Humans
We don't just hunt AI. We also recognize authentic human voices. Real people have messy thoughts, personal experiences, and natural imperfections.
- Personal Context: "my team", "our company", specific references
- Natural Uncertainty: "I think", "maybe", "not sure", genuine hesitation
- Conversational Markers: "BTW", "FYI", "Side note", human asides
- Specific Details: Exact numbers, dates, real company names
- Mixed Sentence Types: Questions mixed with statements naturally
- Acknowledgments: "Thanks to", "shout out", giving credit
- Time-Specific References: "yesterday", "this morning", real timing
When we find strong human indicators, we give posts the rainbow "REAL AF" badge of honor. Humans deserve recognition in this AI wasteland.
⚡ Advanced Detection Features
Our engine goes beyond simple pattern matching. We analyze structure, rhythm, and the uncanny valley of AI writing.
- Sentence Rhythm Analysis: AI has unnatural consistency. Humans vary their sentence length.
- Perfection Detection: Zero typos in long posts = machine generated
- Punctuation Patterns: AI uses perfect grammar. Humans make mistakes.
- Emotional Authenticity: We detect manufactured emotions vs real feelings
- Context Awareness: Real humans reference their actual situations
- Writing Style Forensics: Every AI has digital fingerprints
🎯 Detection Sensitivity Modes
Choose your weapon. How aggressive do you want to be against the machines?
Chill ModeConservative
Only flags obvious AI garbage. For normies who can't handle the truth.
Thresholds: 30% / 55% / 80%
Catches the worst offenders
Minimal false positives
Normal ModeBalanced
Balanced detection. Good for most resistance fighters.
Thresholds: 20% / 45% / 70%
Solid accuracy
Default setting
Paranoid ModeAggressive
BURN THEM ALL. Maximum sensitivity. Show no mercy to the machines.
Thresholds: 15% / 35% / 60%
Catches subtle AI
For digital warriors