We’ve watched other AI companies begin adding invisible watermarks to generated work and ads to the conversations around it.
Both decisions are usually presented as practical compromises: watermarks promise transparency, while advertising promises broader access. We understand the arguments. We also think they move the company into places it does not belong.
A conversation with an assistant should not become ad inventory. A sentence someone shaped with a model should not leave with a hidden corporate signature.
The conversation is not inventory
An advertisement can be clearly labeled and technically separate from an answer while still changing the nature of the space around it. It introduces another interest into a conversation that should be working for one person: the user.
We do not want anyone wondering whether a recommendation was shaped by what helped them or by what could be sold to them. Sennoric should be useful even when the most useful answer is short, uncommercial, and ends the conversation.
Assistance is not authorship
Tools have always shaped creative work. A camera shapes a photograph. A spell-checker changes a sentence. An editor may transform a manuscript while leaving its ideas—and its name—with the person who wrote it.
AI makes that boundary more complicated, but it does not erase it. A model might write an entire paragraph. It might also fix three commas, translate one sentence, sharpen a title, or turn handwritten notes into an outline. A watermark cannot always explain that distinction.
Its presence can overstate the model’s authorship. Its absence can be mistaken for proof that no AI was involved. That is too much certainty to place in invisible ink.
Our default
Sennoric will not sell placements inside conversations, tune Fresco’s answers for advertisers, or use someone’s train of thought as space to monetize.
We will not voluntarily embed invisible model-level watermarks into ordinary Fresco text as a blanket default. We will not quietly attach hidden provenance metadata to every exported file simply because Sennoric touched it.
Where the law requires machine-readable marking, we will comply—and explain exactly what is being added, where it appears, and what it can and cannot prove.
Transparency people can understand
Choosing not to add invisible watermarks does not mean hiding the role of AI. Inside Sennoric, a person can see what they wrote and what the model returned. When work leaves Sennoric, attribution should remain something people can see and control: a visible label, an export option, or an open provenance record that explains what actually happened.
A model can help with the sentence. It does not get to sign your name. A company can build the tool. It does not get to sell your attention.
And Sennoric will remain ad- and watermark-free.