2023-08-30 09:37:41 ET
%Google parent company %Alphabet () says it is testing a new %DigitalWatermark that could potentially identify images created using %ArtificialIntelligence (A.I.)
Alphabet says the watermark is being developed by DeepMind, its A.I. research unit, and is aimed at countering disinformation.
The new watermark would be invisible to the human eye but detectable by computers.
However, Alphabet is warning that the A.I. watermark is likely not foolproof against so called “deep fakes,” images and videos that have been extremely manipulated.
Concerns are growing that the proliferation of A.I. will make it increasingly difficult to distinguish between real images and artificially generated ones.
Watermarks are typically a logo or text added to an image to identify its authenticity and ownership, as well as to make it difficult for an image to be copied or used without permission.
Many technology companies already use a technique called “hashing” that creates digital fingerprints on videos, so they can be tracked online and removed if needed.
Alphabet is warning that it’s A.I. watermark is in the early test phase and that it needs people to use it to learn more about how effective it is with artificial images.
Other technology companies, including %Microsoft () and %Amazon (), have also announced plans to watermark A.I.-generated content.
In China, the government has banned all A.I.-generated images without watermarks. Similar measures are being debated in other countries, including the U.S.
Alphabet’s stock has increased 51% this year and currently trades at $134.57 U.S. per share.