Watermarks & The Fight Against Fakes

Welcome to another edition of the best damn newsletter in AI. Here we’ll break down AI topics that matter, open your mind to use cases, and keep you ahead of the curve.

Our #1 goal is to be useful. So please shoot us an email 📩 if you have questions or feedback, and especially if you implement something we share!

Here's what we're covering today:

• Using watermarks for AI content + the new Frontier Model Forum

• Using AI to write your emails (finally!)

• More AI from non-AI companies

Let’s get to it! 👇

TODAY'S PERSPECTIVE

AI Content Watermarks & The Fight Against Fakes

As AI adoption grows, the authenticity of digital content is becoming a pressing issue. And the White House has recently put a spotlight on the problem.

The primary concern is about something called deep fakes. Deep fakes are videos or images that have been spoofed to look like someone else and can be used to spread misinformation. But overall “fake” content, is a problem. And as AI-generated photos and videos have gotten ridiculously real as of late, the line between real and fake content is blurry and alarming.

With upcoming elections in mind - there's certainly urgency in being able to identify AI-generated content.

7 of these AI leaders have publicly stepped up and started the Frontier Model Forum. They've shared public commitments, echoed in their blog posts, to responsible AI development. (read OpenAI’s, Google’s)

And one item is explicitly on the docket: AI content watermarking.

The concept of AI watermarking isn't new. We talked about it in January.

While the specifics of how they implement it will likely remain under wraps to prevent people from just quickly scrubbing them away - one possible method includes adding a unique statistical pattern to the text generation process of models like ChatGPT. Some research has shown this method could even stay intact with lightweight editing.

There are also more explicit watermarking methods in play. Like Adobe's Firefly and OpenAI's DALL-E, already have a simple icon in the corner of the image—a visible watermark, if you will.

We're happy to see industry leaders committed to ensuring the authenticity of digital content.

And as watermarking content progress, this will certainly have interesting implications for the use of AI in our everyday work. So we’re paying attention.

USE CASE

Finally, emails write themselves

We had the privilege to get early access to Superhuman’s new AI features. And let us tell you - this is the AI email generator we’ve been looking for.

First, we’re fans of Superhuman* and have been personal customers of “the fastest email experience ever made” for years.

Now they’ve added an AI-powered email drafter that actually works.

Our favorite parts 🤩 

1. Superhuman AI learns your tone & voice

2. It learns the context of your role & your business

3. And the email drafts are quite good, actually saving tons of time!

Here’s a peek of how it works:

Superhuman is great for startups, executives, investors and personal email. (Not as good if you’re currently stuck in Outlook or have strict email tool policies at work)

If you want to check it out, Superhuman AI* is now available to all Superhuman customers 🙌 

*This is an affiliate link, and we only recommend tools that we personally vet and will stand behind.

LINKS

For your reading list 📚

AI adoption continues to spread outside of AI companies (the good & the bad)…

• Stack Overflow’s traffic is down 50% now, a company that people thought would be disrupted by engineers using ChatGPT… is looking to be true

• Wayfair is experimenting with AI. The home goods company has a new feature that will let you redraw your living room

And if you’re really nerdy…  

• GitHub, Hugging Face, and others are trying to guard open source development of AI in the new EU AI Act

That's all!

We'll see you again on Tuesday. Thoughts, feedback and questions are much appreciated - respond here or shoot us a note at [email protected].

Cheers,

🪄 The AI Exchange Team