Real vs. AI-Made — Can You Tell?

AI pictures look real — sometimes too real. Here's how to train your eyes to spot the clues that tell you 'a machine made this.'

18 min · Reviewed 2026

A new kind of puzzle

Just a few years ago, you could tell AI pictures instantly — eyes looked weird, hands had seven fingers, text was gibberish. In 2026, it's way harder. Some AI pictures fool even grown-ups. But there are still clues.

The detective's checklist

Count the fingers. AI sometimes draws 4, 6, or weirdly bent fingers.
Look at the background. Is a street sign gibberish? Are letters half-wrong?
Check ears, teeth, and earrings — weirdly asymmetric pairs are a clue.
Look at shadows. Do they all go the same way? Real photos do.
Zoom in on hair. AI sometimes makes hair flow into the background in impossible ways.
Check reflections in eyes, water, or glass — AI often gets these subtly wrong.

Real photo clues	AI picture clues
Messy, imperfect details everywhere.	Skin and hair look almost TOO smooth.
Text in the background reads correctly.	Background text is squiggly or wrong.
Shadows follow one light source.	Shadows sometimes fight each other.
Faces have real asymmetry.	Faces feel uncanny — too symmetric.

Content Credentials — a little tag that helps

Some pictures now come with a tiny invisible tag called Content Credentials (made by a group called C2PA). If you see a small 'CR' badge on an image, click it — it tells you what tool made the picture and when. Adobe, Microsoft, the BBC, and many news sites use it.

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-creative-real-vs-ai-explorers

What is the core idea behind "Real vs. AI-Made — Can You Tell?"?
1. AI pictures look real — sometimes too real. Here's how to train your eyes to spot the clues that tell you 'a machine made this.'
2. 'Suggest 3 funny code names for objects in my room.'
3. 'Help me design a magic cape with one special power.'
4. Version. Tag the LoRA with training date, base model, and trigger token.
Which term best describes a foundational idea in "Real vs. AI-Made — Can You Tell?"?
1. Content Credentials
2. AI detection
3. C2PA
4. 'Suggest 3 funny code names for objects in my room.'
A learner studying Real vs. AI-Made — Can You Tell? would need to understand which concept?
1. AI detection
2. C2PA
3. Content Credentials
4. 'Suggest 3 funny code names for objects in my room.'
Which of these is directly relevant to Real vs. AI-Made — Can You Tell??
1. AI detection
2. Content Credentials
3. 'Suggest 3 funny code names for objects in my room.'
4. C2PA
Which of the following is a key point about Real vs. AI-Made — Can You Tell??
1. Count the fingers. AI sometimes draws 4, 6, or weirdly bent fingers.
2. Look at the background. Is a street sign gibberish? Are letters half-wrong?
3. Check ears, teeth, and earrings — weirdly asymmetric pairs are a clue.
4. Look at shadows. Do they all go the same way? Real photos do.
Which of these does NOT belong in a discussion of Real vs. AI-Made — Can You Tell??
1. Check ears, teeth, and earrings — weirdly asymmetric pairs are a clue.
2. Count the fingers. AI sometimes draws 4, 6, or weirdly bent fingers.
3. Look at the background. Is a street sign gibberish? Are letters half-wrong?
4. 'Suggest 3 funny code names for objects in my room.'
What is the key insight about "Sometimes you just can't tell" in the context of Real vs. AI-Made — Can You Tell??
1. 'Suggest 3 funny code names for objects in my room.'
2. 'Help me design a magic cape with one special power.'
3. That's the truth. Top AI images in 2026 can fool experts. That's why the rule is: don't trust a picture alone.
4. Version. Tag the LoRA with training date, base model, and trigger token.
What is the key insight about "A world-champion skill" in the context of Real vs. AI-Made — Can You Tell??
1. 'Suggest 3 funny code names for objects in my room.'
2. 'Help me design a magic cape with one special power.'
3. Version. Tag the LoRA with training date, base model, and trigger token.
4. Learning to pause before you believe a picture is one of the most important 2020s skills.
What is the recommended tip about "Make it your own!" in the context of Real vs. AI-Made — Can You Tell??
1. AI gives you a starting point — you add the magic. Take what AI generates and change at least 3 things to make it truly …
2. 'Suggest 3 funny code names for objects in my room.'
3. 'Help me design a magic cape with one special power.'
4. Version. Tag the LoRA with training date, base model, and trigger token.
Which statement accurately describes an aspect of Real vs. AI-Made — Can You Tell??
1. 'Suggest 3 funny code names for objects in my room.'
2. Just a few years ago, you could tell AI pictures instantly — eyes looked weird, hands had seven fingers, text was gibberish.
3. 'Help me design a magic cape with one special power.'
4. Version. Tag the LoRA with training date, base model, and trigger token.
What does working with Real vs. AI-Made — Can You Tell? typically involve?
1. 'Suggest 3 funny code names for objects in my room.'
2. 'Help me design a magic cape with one special power.'
3. Some pictures now come with a tiny invisible tag called Content Credentials (made by a group called C2PA).
4. Version. Tag the LoRA with training date, base model, and trigger token.
Which best describes the scope of "Real vs. AI-Made — Can You Tell?"?
1. It is unrelated to creative workflows
2. It applies only to the opposite professional tier
3. It was deprecated in 2024 and no longer relevant
4. It focuses on AI pictures look real — sometimes too real. Here's how to train your eyes to spot the clues that tel
Which section heading best belongs in a lesson about Real vs. AI-Made — Can You Tell??
1. The detective's checklist
2. 'Suggest 3 funny code names for objects in my room.'
3. 'Help me design a magic cape with one special power.'
4. Version. Tag the LoRA with training date, base model, and trigger token.
Which section heading best belongs in a lesson about Real vs. AI-Made — Can You Tell??
1. 'Suggest 3 funny code names for objects in my room.'
2. Content Credentials — a little tag that helps
3. 'Help me design a magic cape with one special power.'
4. Version. Tag the LoRA with training date, base model, and trigger token.
Which of the following is a concept covered in Real vs. AI-Made — Can You Tell??
1. Content Credentials
2. C2PA
3. AI detection
4. 'Suggest 3 funny code names for objects in my room.'

← Back to interactive lesson

Tendril · Explorers · Creative AI

Real vs. AI-Made — Can You Tell?

AI pictures look real — sometimes too real. Here's how to train your eyes to spot the clues that tell you 'a machine made this.'

18 min · Reviewed 2026

A new kind of puzzle

The detective's checklist

Count the fingers. AI sometimes draws 4, 6, or weirdly bent fingers.
Look at the background. Is a street sign gibberish? Are letters half-wrong?
Check ears, teeth, and earrings — weirdly asymmetric pairs are a clue.
Look at shadows. Do they all go the same way? Real photos do.
Zoom in on hair. AI sometimes makes hair flow into the background in impossible ways.
Check reflections in eyes, water, or glass — AI often gets these subtly wrong.

Real photo clues	AI picture clues
Messy, imperfect details everywhere.	Skin and hair look almost TOO smooth.
Text in the background reads correctly.	Background text is squiggly or wrong.
Shadows follow one light source.	Shadows sometimes fight each other.
Faces have real asymmetry.	Faces feel uncanny — too symmetric.

Content Credentials — a little tag that helps

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-creative-real-vs-ai-explorers

What is the core idea behind "Real vs. AI-Made — Can You Tell?"?
1. AI pictures look real — sometimes too real. Here's how to train your eyes to spot the clues that tell you 'a machine made this.'
2. 'Suggest 3 funny code names for objects in my room.'
3. 'Help me design a magic cape with one special power.'
4. Version. Tag the LoRA with training date, base model, and trigger token.
Which term best describes a foundational idea in "Real vs. AI-Made — Can You Tell?"?
1. Content Credentials
2. AI detection
3. C2PA
4. 'Suggest 3 funny code names for objects in my room.'
A learner studying Real vs. AI-Made — Can You Tell? would need to understand which concept?
1. AI detection
2. C2PA
3. Content Credentials
4. 'Suggest 3 funny code names for objects in my room.'
Which of these is directly relevant to Real vs. AI-Made — Can You Tell??
1. AI detection
2. Content Credentials
3. 'Suggest 3 funny code names for objects in my room.'
4. C2PA
Which of the following is a key point about Real vs. AI-Made — Can You Tell??
1. Count the fingers. AI sometimes draws 4, 6, or weirdly bent fingers.
2. Look at the background. Is a street sign gibberish? Are letters half-wrong?
3. Check ears, teeth, and earrings — weirdly asymmetric pairs are a clue.
4. Look at shadows. Do they all go the same way? Real photos do.
Which of these does NOT belong in a discussion of Real vs. AI-Made — Can You Tell??
1. Check ears, teeth, and earrings — weirdly asymmetric pairs are a clue.
2. Count the fingers. AI sometimes draws 4, 6, or weirdly bent fingers.
3. Look at the background. Is a street sign gibberish? Are letters half-wrong?
4. 'Suggest 3 funny code names for objects in my room.'
What is the key insight about "Sometimes you just can't tell" in the context of Real vs. AI-Made — Can You Tell??
1. 'Suggest 3 funny code names for objects in my room.'
2. 'Help me design a magic cape with one special power.'
3. That's the truth. Top AI images in 2026 can fool experts. That's why the rule is: don't trust a picture alone.
4. Version. Tag the LoRA with training date, base model, and trigger token.
What is the key insight about "A world-champion skill" in the context of Real vs. AI-Made — Can You Tell??
1. 'Suggest 3 funny code names for objects in my room.'
2. 'Help me design a magic cape with one special power.'
3. Version. Tag the LoRA with training date, base model, and trigger token.
4. Learning to pause before you believe a picture is one of the most important 2020s skills.
What is the recommended tip about "Make it your own!" in the context of Real vs. AI-Made — Can You Tell??
1. AI gives you a starting point — you add the magic. Take what AI generates and change at least 3 things to make it truly …
2. 'Suggest 3 funny code names for objects in my room.'
3. 'Help me design a magic cape with one special power.'
4. Version. Tag the LoRA with training date, base model, and trigger token.
Which statement accurately describes an aspect of Real vs. AI-Made — Can You Tell??
1. 'Suggest 3 funny code names for objects in my room.'
2. Just a few years ago, you could tell AI pictures instantly — eyes looked weird, hands had seven fingers, text was gibberish.
3. 'Help me design a magic cape with one special power.'
4. Version. Tag the LoRA with training date, base model, and trigger token.
What does working with Real vs. AI-Made — Can You Tell? typically involve?
1. 'Suggest 3 funny code names for objects in my room.'
2. 'Help me design a magic cape with one special power.'
3. Some pictures now come with a tiny invisible tag called Content Credentials (made by a group called C2PA).
4. Version. Tag the LoRA with training date, base model, and trigger token.
Which best describes the scope of "Real vs. AI-Made — Can You Tell?"?
1. It is unrelated to creative workflows
2. It applies only to the opposite professional tier
3. It was deprecated in 2024 and no longer relevant
4. It focuses on AI pictures look real — sometimes too real. Here's how to train your eyes to spot the clues that tel
Which section heading best belongs in a lesson about Real vs. AI-Made — Can You Tell??
1. The detective's checklist
2. 'Suggest 3 funny code names for objects in my room.'
3. 'Help me design a magic cape with one special power.'
4. Version. Tag the LoRA with training date, base model, and trigger token.
Which section heading best belongs in a lesson about Real vs. AI-Made — Can You Tell??
1. 'Suggest 3 funny code names for objects in my room.'
2. Content Credentials — a little tag that helps
3. 'Help me design a magic cape with one special power.'
4. Version. Tag the LoRA with training date, base model, and trigger token.
Which of the following is a concept covered in Real vs. AI-Made — Can You Tell??
1. Content Credentials
2. C2PA
3. AI detection
4. 'Suggest 3 funny code names for objects in my room.'

← Back to interactive lesson