How it works
How Lykanor assesses content.
Lykanor does not answer the question “true or false?” with a single verdict. Instead, we gather signals from several independent directions and make transparent what they mean – and what they do not.
1. Understanding the input
It starts with the content: a link, an image, a video, an audio file or a text. Lykanor first examines what is actually there – format, origin and basic technical information.
2. Technical signals & provenance
Metadata and provenance information can provide first hints: when and how was a file created or modified? If such information is missing, that too is a relevant signal – but not proof on its own.
3. Media analysis
The content itself is examined for anomalies that may point to AI generation or manipulation – lighting and detail in images, voice characteristics in audio, style and structure patterns in text.
4. Sources & context
Especially for verifiable claims, looking outward is crucial: what do independent sources report? Does the content appear in a different context? Has it already been assessed or debunked?
5. Combining the evidence
All signals are collected and weighed against each other. Where signals contradict, that is made visible – not smoothed away.
6. An understandable result
The result is an assessment with an explanation: which signals exist, how strong they are, which sources were consulted – and where the limits of the statement lie.
Why uncertainty is part of it
No method can reliably detect every fake – and anyone promising otherwise creates false trust. When the evidence is insufficient, Lykanor shows exactly that: an unclear situation, openly named. That is not a flaw, it is the core of an honest result.
And when it needs to go deeper?
For complex cases, a deeper analysis layer is planned: more sources, timelines and structured casework. It builds on the same evidence logic – just more thoroughly.
Want to be there at launch?
Join the beta list – we will get in touch as soon as the analysis opens.