Abstract: A method for operating a functionally safe audio output system with a computing node with a simple certification and a complex computing node, the method comprising generating a code assigned to a warning in the computing node with a simple certification, transmitting the code to the complex computing node, generating, in a first computing process, an output signal having an acoustic watermark, transmitting the output signal to a second computing process independent of the first computing process, determining, in the second computing process independent of the first computing process, a code contained in the output signal, transmitting the code to the computing node with a simple certification, and verifying whether the audio output system is functionally safe on the basis of a comparison of the generated code with a code based on the received code.