01
Sourced
A finding needs a citable, checkable origin: platform-confirmed data, direct first-party evidence with a preserved chain of custody, or another checkable source. Unsourced claims are not carried forward as findings.
Methodology
What counts as verified, and why it matters.
Detection technology can flag something as suspicious. It doesn't tell you whether a platform, a court or a newsroom will act on it. That gap is a documentation and verification question: this is the standard Eterna applies to close it.
The four-part standard
A finding is marked verified only when it clears all four.
02
Corroborated
A single unverified signal is logged as reported, not verified, until an independent second signal supports it. One screenshot or one report alone does not clear the bar.
03
Attributable to a method
Every finding states how the determination was made (forensic review, platform confirmation, subject confirmation or investigative review), not just the conclusion.
04
Authorized before enforcement
Where a finding leads to an active response, action requires confirmed authorization from the affected party or their authorized representative before anything is escalated.
What doesn't clear the bar
Reported, not verified, and shown that way.
Anything that doesn't pass all four checks is labeled visibly as reported but unconfirmed or disputed, not quietly dropped. A record that only shows confirmed outcomes isn't a verification standard, it's marketing.
This is also why detection accuracy alone is not treated as protection. Independent benchmark testing has found automated detector accuracy can fall substantially outside lab conditions compared to curated test sets, and human visual identification of high-quality synthetic video has been measured well below reliable levels in published research. A confidence score is one input. It is not a case.
Where this standard is applied
Case review, and the Identity Response Observatory.
This is the same standard used inside client case review, and the standard behind every entry considered for the Eterna Identity Response Observatory: Eterna's public record of documented digital-identity incidents.
This same standard is applied across Eterna's deepfake protection, AI impersonation protection and online reputation protection work.