Thursday Thoughts on AI

Thursday Thoughts on AI: Artificial Intelligence and Attribution

Written by Mark Fallon | Aug 27, 2026, 1:30:01 PM

When presenting content to an employer, a client, or the public, an honorable person acknowledges any work created by someone else. Citations, footnotes, bibliographies, and endnotes are just some of the ways authors identify sources. When posting someone else’s photos or artwork, the creator is credited.

When someone uses editing tools on their own work, there is usually no mention of the software. This blog is written and edited using Microsoft Word, including accepting some autofill text and grammar corrections. Almost every image was altered in Adobe Photoshop. That information isn’t normally posted.

The widespread use of artificial intelligence (AI) to generate or manipulate content – text, images, video – has created a new challenge. When should the use of AI be acknowledged and when is it “just software”? Does it make a difference if the AI agent generates content that is then edited by the user, or if the user creates content that is manipulated by the AI agent?

In the midst of this discussion, Anthropic announced that they are now adding a digital watermark to all work generated or processed on their Claude platform. This policy is aligned with the agreement signed with the European Union (EU) and the EU AI Act's Article 50(2) Code of Practice on Transparency of AI-Generated Content.

The watermark isn’t visible to the human eye, but embedded in the metadata of the file. Anthropic admits that this only means that the content was processed by Claude – it doesn’t confirm the original author. If you asked Claude to edit an original image, the watermark is added. In some cases the AI-content can be manipulated in a way to remove the watermark.

The response to the announcement has been mixed. It’s sparked a much-needed debate about creativity, originality, tools, and proper attribution. There’s no consensus, and probably won’t be for some time.

Amazing Astronomical Fact: For over 2,500 years, humans have been able to predict solar eclipses. By tracking and recording celestial events, ancient astronomers detected recurring patterns and created models to predict the future. Savvy leaders could use that information, combined with the mythology attached to an eclipse, to manipulate the populace.

While scientific knowledge has replaced most myths, people remain fascinated by eclipses. The August 21, 2017 was commemorated by the US Postal Service with a special forever stamp. When the image of the total solar eclipse was touched, the heat of your finger would transform it into an image of the Moon.

The path of totality for the 2026 eclipse crossed northern Russia, the Arctic, eastern Greenland, western Iceland, northeastern Portugal, and northern Spain. Watch parties were held across Europe. In the United States, a partial eclipse was visible from Alaska, the Great Lakes Region, and the Northeast/Mid-Atlantic area.

#artificialintelligence #ai #watermark #attribution #solareclipse #eclipse #theberkshirecompany

Missed a past newsletter? No worries they're all saved on our Thursday Thoughts on AI page.