Microsoft’s New Pledge: Watermarking AI-Generated Content – A Step Towards Transparency, Or Not Enough?

car parked in front of building

Greetings from PTTOW! in San Diego. I’m here to keynote about AI, then help facilitate a town hall meeting about its cultural impact. Generative AI raises a bunch of questions about copyright, ownership, human creativity, and content provenance.

In line with these questions, Microsoft (MSFT) is taking a proactive stance. As the lines between reality and AI-created content continue to blur, the tech giant is stepping up with a new initiative: watermarking AI-generated images and videos. At its annual Build conference, Microsoft announced upcoming features for its Bing Image Creator and Designer, which will allow consumers to verify whether content was AI-created.

This isn’t as simple as slapping on a visible watermark. The new system uses cryptographic methods to sign AI-generated content with metadata about its origin, but there’s a hitch; to read this signature, platforms will need to adopt the Coalition for Content Provenance and Authenticity (C2PA) interoperable specification, a standard developed with input from industry heavyweights including Adobe, Arm, Intel, and Truepic.

Microsoft’s move aligns with broader industry trends. As generative AI takes hold, other companies are also adopting mechanisms to track media provenance. Google has started using embedded metadata to signal AI-created visual media, while Shutterstock (along with AI startup Midjourney) has implemented guidelines to indicate when content was produced by a generative AI tool.

Will these efforts be enough in an industry where not all players have embraced similar standards? Will Microsoft’s initiative be the catalyst to widespread adoption, or is it merely a drop in the ocean? The answer depends on whether other industry giants follow suit, and whether consumers demand more transparency in the content they consume.

We stand at an interesting crossroads, where technology’s incredible capabilities can sometimes outpace our capacity to comprehend or regulate them. The next few months will be telling as we see how these watermarking features are received and whether they spur wider change within the industry.


More By This Author:

OpenAI Proposes International Regulatory Body for AI
OpenAI Introduces ChatGPT App For iOS
Google I/O 2023

Disclosure: This is not a sponsored post. I am the author of this article and it expresses my own opinions. I am not, nor is my company, receiving compensation for it.

How did you like this article? Let us know so we can better customize your reading experience.

Comments

Leave a comment to automatically be entered into our contest to win a free Echo Show.
Or Sign in with