THE SHORT ANSWER

Synthetic media includes AI-generated or manipulated images, audio and video. Appearance, voice and realism can support a claim but cannot authenticate a person or event alone. For important decisions, confirm through a known channel and seek independent evidence.

Synthetic media changes the cost of convincing signals

Images can depict events that did not occur, audio can resemble a familiar speaker, and video can alter a person's appearance or speech. Manipulation predates generative AI; current tools can make production faster and more accessible.

Do not rely on a list of visual glitches. Generation and detection methods change, compression creates false clues, and authentic media can look imperfect.

Evidence & context: NIST

Verify the claim, identity and requested action

  • Claim: what exactly is the media asking you to believe?
  • Identity: can the person be reached through a known contact route?
  • Context: is there independent reporting, an original source or a complete recording?
  • Action: what happens if you wait before sending money, data or access?

Corroboration is stronger than confidence

Look for evidence created independently of the disputed media. Compare the claim with primary sources and reliable reporting. How to Compare Conflicting Claims provides the deeper evidence method.

Do not amplify uncertain material

Pause forwarding, preserve the original location if safe, report impersonation through the relevant platform or organization, and warn the person being impersonated through another channel. Label uncertainty clearly while facts are checked.

Sources & further reading

  1. How To Avoid a Government Impersonation Scam

    United States Federal Trade Commission. Official consumer guidance on urgency, payment demands, caller-ID limits and contacting an organization through a known channel. Agency examples and reporting routes are United States-specific.

  2. NIST SP 800-63-4: Digital Identity Guidelines

    National Institute of Standards and Technology. The 2025 guideline distinguishes identity proofing, authentication and federation and selects assurance according to risk. It is written for United States federal systems but offers a useful conceptual reference beyond them.

  3. Generative Artificial Intelligence Profile (NIST AI 600-1)

    NIST. Risk-management guidance, including confabulation. It does not establish a universal error rate.

Examples and exercises are illustrative unless attributed to a source. No independent expert review is claimed.

A correction, a counterexample or an experience worth sharing?

Join the conversation ↗