Deepfakes: when fraud meets the crisis of evidence

Voice cloning is now one of the most worrying uses of deepfakes for businesses.

Just a few seconds of a public address can be enough to replicate an executive’s voice, impersonate them and carry out fraud: issuing a bank transfer order, disclosing sensitive information, manipulating an employee…

Our partner Claire Poirson appeared on the 𝗖𝗧𝗥𝗟+𝗦(𝗮𝗳𝗲) podcast, presented by Philippine Dolbeau for ESET France, to analyse the legal implications of this development.

Beyond fraud, deepfakes call into question our ability to assess the authenticity of evidence

Images, audio and video have long been particularly persuasive pieces of evidence in legal proceedings.

What happens when their authenticity itself becomes questionable?

A deepfake has already been invoked as a defence before a French court. With the rapid advancement of generative technologies, identifying manipulation could become much more complex in the future, with direct consequences for the handling and assessment of evidence.

For businesses, therefore, the risk must be addressed at an early stage

Audit procedures, cyber security, crisis management mechanisms and internal validation processes must now take into account the possibility of AI-enabled fraud.

The law is gradually addressing these practices, notably through the SREN Act and the European regulation on artificial intelligence (AI Act).

However, regulation alone cannot prevent fraud that relies precisely on the ability to artificially create trust.

When it comes to deepfakes, the primary challenge is therefore as much a legal one as it is a practical one: to stop regarding as authentic anything that merely appears plausible.

Claire Poirson