How can I verify human authorship in digital content?
Lately I’ve seen a number of people on LinkedIn complaining about those of us that ask, is this AI or human originated? They comment with a bit of disdain that it does not matter and nobody cares. Well somebody cares and the Supreme Court agrees that identifying human authorship is necessary. In fact, so necessary, it is the bedrock of copyright law. Restated a different way, digital content created by artificial intelligence (AI) has no copyright protection.
Wait, what did you say? Digital content does not have copyright protection when it is created by AI. This little Supreme Court Opinion should make every marketing executive, business owner, digital creator or business owner pay serious attention to whether their content is created and a live human or artificial intelligence.
Think about it, all of your copy, images and video are free game for anyone to use anywhere. It’s not protected if it’s created by AI. That new logo that you love so much, yep, everyone anyone can use it. Your new web site design, yep, fair game. How about the brand guidelines describing your logo, tagline and soundtrack? Yep, those too!
So how do you verify human authorship in digital content? First and foremast, ask if it is created by a human or by AI. If your copywriter or agency is not willing to attest that they created it themselves and did not use AI, you should worry.
I recently asked a CEO advisor if his LinkedIn post was written by AI. His response? Silence. He wasn’t about to give up his position and confirm that his writing was not really his. Sure, it was his idea and his thoughts and submitted a prompt to his AI engine of choice – but the writing, the tone, content and creativity was not him. It was artificial intelligence. The post did not reflect his intelligence – no, it was artificial and deceptive because he could now own up to the fact it was written by someone, or in this case, something else.
Separately I spoke with the Founder and CEO of advertising platform startup today. She readily admitted that the first LinkedIn message sent to me was entirely AI generated. All subsequent responses between us were hers, and hers alone. It was human originated writing – human authorship. I ran her initial message through our AI detection software, AI TruthTeller® and confirmed the same. The first message was AI generated. In actuality, I ran the text message first and then asked her to confirm if it was AI. This is a simple example of using a human-in-the-loop (HITL) approach to verifying human authorship. Clearly this requires honest feedback and a willingness to disclose authorship, but that is the reality we now live in. A human-in-the-loop process is one way to verify human authorship of digital content. Ask and verify and if necessary put it in writing.
Another method that has merit, but is not fool proof is to use AI detection solutions like AI TruthTeller. Sadly many of the largest AI detectors focus only or primarily on text and by that I mean long-form text. These are web based solutions that offer analysis of text based writings. Some of them require 300 words or more to function properly while others may provide an answer on less, they are unpredictable. AI TruthTeller is built on conversations between people, so there is no minimum word count requirement. Clearly, the more words the better, but that LinkedIn message I mentioned earlier, it was 51 words. Like all LLM derived solutions, model training is required and the more people and messages that are processed, the better the outcomes.
One other point in answer to the question, how do I verify human authorship of digital content? Blended content is particularly difficult to detect. This is true of the written word as well as images.
We recently ran a test of 300 photographic images. These were all original pictures, created by us on devices we own. This allowed us to know exactly where they came from and to know they were created by traditional photographic devices. What was the outcome? 99% correct. AI TruthTellergot it wrong on 3 pictures. How is this possible? Well, in our algorithms, we embrace digital forensic practices, provenance signatures and chain of custody. In short, whenever possible we focus on the device not the output. Generative AI creates an infinite number of pixels, in real time, across a multitude of platforms and devices. However, when we can get to device level details, we use them and that increases accuracy. Sure, web platforms strip off device details and AI is increasingly embedded in mobile devices, but for now we’ll take what we can get and will continue to embrace forensic details from the original device.
So, back to the question, how do I verify human authorship of digital content? There is no single simple answer. Asking your creative providers is a good start.Getting signed affidavits is a step forward. Setting production standards for employees or contractors and auditing device level originals adds another layer. Searching for provenance signatures is another good step, and the list goes on.
Verifying human authorship is a necessary part of today’s world. Ignore the critics that complain when you ask. If you want copyright protection and proof of ownership, now is the time to embrace all of the above methods. Your attorneys and your customers will thank you.
Trust is critically important and if the basic assumption in the market is that everything is touched by AI, edited, enhanced or originally generated by AI, brand assets, brand strength and brand trust will quickly vanish.
Noted legal opinions:
USC Libraries: Human Authorship and Creative Control
United States Court of Appeal; Thaler vs. Permutter, Director of the United States Copyright Office: