SecMedID – Secure Medical Identities

Detecting deepfakes and fake identities

Abstrakte Grafik von sechs polygonalen Gesichtern in Rot- und Blautönen, die durch KI-Deepfakes erstellte, gefälschte Identitäten symbolisieren. Das Bild dient als visueller Header für das BSI-Projekt „SecMedID“ zur Erkennung medialer Fälschungen und Abwehr von Desinformation.

Using deepfake technology, the faces or voices of people appearing in videos can be automatically replaced, often with startling realism.

Fake identities created in this way pose a number of risks:

  • New forms of fraud and extortion targeting individuals and organizations
  • Loss of trust in the press and broadcasting
  • Influencing political processes through fake news and disinformation
  • Discrediting of individuals through defamatory videos
  • Fabrication of evidence in criminal and civil proceedings
  • Threats to internal and external security through false reports to the police, the general public, or the armed forces.

This makes it necessary to be able to reliably verify the authenticity of individuals in digital media (their “media identity”). To this end, the “SecMedID” project will address both AI-powered deepfake attack tools used to fabricate media identities and defensive measures against them.

The “SecMedID” project was commissioned by the Federal Office for Information Security (BSI). Duration: June 2023 to February 2025.

Contact
More Info
Link to IT-ForensikLink to Media SecurityLink to BSI project page

The SecMEdID project is investigating

  • The state of the art in deepfake methods for video and audio
  • The further development of deepfake methods—face swapping, facial reenactment, voice conversion, and text-to-speech—to assess future risk potentials
  • Forensic detection methods for deepfakes in digital image and audio data, even when obscured by adversarial attacks