Sony’s fight with deepfakes: Working on in-camera authenticator to curb AI-generated fake images

Sony is set to take on AI-generated images through its upcoming ‘in-camera authenticity’ technology, a collaborative effort with non-profit news organization Associated Press (AP) and Camera Bits.

The company has recently announced the completion of the second round of testing for this innovative solution.

This cutting-edge technology introduces a digital signature, essentially creating a birth certificate for images that validate their origin.

Sony aims to address concerns about the authenticity of digital content by generating this signature in the camera’s hardware chipset at the moment a photographer captures an image.

The process is expected to provide professionals with a means to verify the legitimacy of their content, offering an additional layer of security for news agencies.

Neal Manowitz, the COO and President of Sony Electronics, expressed optimism about the valuable results obtained through the testing phase but remained tight-lipped about the intricate workings of the technology.

It is speculated that the images captured using this technology will contain substantial metadata, including details such as the camera model, the time the picture was taken, and any subsequent edits made.

Sony plans to roll out the digital signature feature in 2024 through a firmware update for its Alpha 9 III, Alpha 1, and Alpha 7S III models.

While Sony is at the forefront of this endeavour, it is worth noting that other companies have also delved into developing technologies to detect image tampering.

Leica, for instance, unveiled the M11-P rangefinder in October of this year, incorporating a Content Credential label containing information about the photo’s capture time and any alterations made.

However, the Leica M11-P comes with a hefty price tag of $9000, significantly more expensive than Sony’s Alpha models.

(With input from agencies)

Sony’s fight with deepfakes: Working on in-camera authenticator to curb AI-generated fake imagesRead More

Leave a Reply

Your email address will not be published. Required fields are marked *