Deepfake Detective

Good reasons to verify:

Find the original source — Photos get reposted constantly. Finding who originally posted it tells you a lot. Did it come from a news outlet? A social media account? Someone claiming to be there?

Check for independent reporting — If something significant happened today, local news should be covering it. If no news outlet is reporting it, that's a strong clue.

Look at details — This is useful, but not as proof. Strange details might mean the image is AI-generated, but they might also mean poor lighting, a camera glitch, or something that just looks odd to you.

Sharing because it spread widely — This is a common trap. Things spread on social media regardless of whether they're true.

The key question: Can you find independent confirmation? If not, wait.

Good reasons to verify:

Find the full original video — Short clips without context are especially risky. The context before or after the clip might completely change the meaning.

Check official accounts — If it's a real statement, the person or their official channels should have it. If they deny it or clarify it, that's important information.

Look for news coverage — Major statements by public figures get reported. If this is real and significant, news outlets should be covering it.

Sharing without verification — Even if it turns out to be real, sharing unverified clips spreads incomplete information and fuels confusion.

The key point: Context matters. A short clip without the full story is inherently unreliable.

Here's the nuance:

Visual flaws as proof — This is a common trap. Hands or text might be distorted because of compression, camera angles, poor lighting, or photo editing. Visual flaws are a clue worth investigating, but they're not proof.

Use visual flaws as one clue — This is better. Strange details can be worth noting, but they should be combined with source checks and corroboration.

Assuming it's real because it looks mostly normal — This is also incomplete. Realistic appearance doesn't prove anything either.

The key insight: AI detection from visual clues alone is unreliable. Check the story, not just the image.

Why appearance doesn't tell the whole story:

Trust it because it looks real — This misses the point. Both AI-generated and human-created images can look realistic. Appearance alone isn't evidence.

Check the source and context — This is the right approach. Regardless of what the image looks like, verify where it came from, who posted it, and what independent sources say about the claim.

AI detector tools — These exist, but they're not reliable. They give false positives and false negatives. They're one tool among many, not definitive proof.

The bigger point: AI-generated doesn't automatically mean false. Human-created doesn't automatically mean true. Check the actual claim, not just the medium.