5 · Spotting deepfakes and synthetic media

AI can now generate realistic images, video, and voice. "Deepfakes" can put words in someone's mouth or fabricate events. Treat surprising media as unverified until confirmed.

Practical defenses:

• Consider the source and motive. Who posted it, and what do they want you to feel or do? Strong emotional content spreads before it's checked.
• Look for tells: odd hands/teeth/ears, warped text or backgrounds, mismatched lighting, unnatural blinking or lip-sync. (Tells fade as tech improves, so don't rely on them alone.)
• Check provenance. Content Credentials (C2PA) attach a tamper-evident history to media: who made it and whether AI was involved. Look for that "CR" / Content Credentials marker.
• Trace it. Reverse-image-search; find the original; see if reputable outlets report it.
• Slow down. The single most effective habit is to not share until you've confirmed.

Government agencies now warn that synthetic media is a real threat for fraud and disinformation (CISA et al., 2023). Healthy skepticism is a skill, not cynicism.

Sources

• Coalition for Content Provenance and Authenticity (C2PA). Content Credentials specification. https://c2pa.org
• CISA, NSA, FBI. (2023). Contextualizing Deepfake Threats to Organizations. https://www.cisa.gov/resources-tools/resources/contextualizing-deepfake-threats-organizations