Ethical question as it pertains to AI documents on Word

[rickandcindy23]

I started to type instructions for my niece to type a deed for the timeshare she is buying from a friend. I advised her to type the deed, rather than pay someone for it. She said she would have no idea how to do it. I started a doc on Word, and it asked me if I wanted copilot to help with the deed. I chose the option. "It" wrote an accurate document without my help on how to type a deed.

If AI is getting this information from various sources, is it not plagiarizing documents from whatever sources it's pulling that information? Is that really ethical? How does a program know so much without taking the ideas and thoughts from others.

 

[TravelJoy]

I started to type instructions for my niece to type a deed for the timeshare she is buying from a friend. I advised her to type the deed, rather than pay someone for it. She said she would have no idea how to do it. I started a doc on Word, and it asked me if I wanted copilot to help with the deed. I chose the option. "It" wrote an accurate document without my help on how to type a deed.

If AI is getting this information from various sources, is it not plagiarizing documents from whatever sources it's pulling that information? Is that really ethical? How does a program know so much without taking the ideas and thoughts from others.

The short answer is yes.

This is the premise of LLMs, Large Language Models. Sometimes the companies will pay some money to say Reddit to use its users interactions to train AI, others such as Meta, GOOGLE etc..use their existing users without their consent or have it as a default buried in their t&c, other times they will "scrape"/plagiarize from a variety of sources to train AI models such as what you mentioned not limited to anything and everything on the Internet including news pieces. As you can tell it really is the wild West when it comes to these issues. Artists, writers, newspapers etc...nobody is spared. There are a plethora of lawsuits out there but in the end they will get away with it as it's already too forgone.

 

[rickandcindy23]

I find this quite despicable.

By the way, I paid $668 with escrow for my last deed recording through a company. I am NEVER doing that again. I will type deeds from now on. There is no reason not to do it myself. I have been burned too many times. End of rant!

 

[dioxide45]

It isn't like deeds are copyrighted pieces of work. They are pretty standard legal documents. No issues here.

 

[rickandcindy23]

It isn't like deeds are copyrighted pieces of work. They are pretty standard legal documents. No issues here.

It was the instructions on how to type a deed. It's from somewhere.

Another thing I tried was, "Choosing the right timeshare."

The information came from where? Some probably came from TUG documents.

 

[TravelJoy]

Its going to get much more worst as laws are currently in the works so that anything AI related becomes nearly invincible.

 

[TravelJoy]

It isn't like deeds are copyrighted pieces of work. They are pretty standard legal documents. No issues here.

I'm inclined to agree with this. It's no different then making your own will. While legal may want to charge $$$ for it, you can well draft your own with the assistance of free programs and simply have it notarized for free in many instances.

 

[Brett]

It isn't like deeds are copyrighted pieces of work. They are pretty standard legal documents. No issues here.

yes, we've use standard "fillable PDF" legal forms for a number of things

 

[rickandcindy23]

I still think it's unethical for AI to search out documents for whatever subject and fill in the information.

AI is scary.

 

[WaikikiFirst]

If AI is getting this information from various sources, is it not plagiarizing documents

One of my 1st reactions to these chatbots was to nickname them "The Universal Plagiarizer". What you're worried about isn't the #1 "ethical" problem with them, and I wouldn't say "ethics" is the #1 problem with them overall. btw, I just saw a few headlines ...

 

[WaikikiFirst]

8:24 AM, 09/15/2025
Rolling Stone publisher Penske Media is suing Google over AI-scraped summaries that appear at the top of web pages (searches, they mean, I guess). The AI-summaries that pop up when using Google's search ... in a potentially landmark lawsuit.

They are damn useful. THey have almost eliminated the plunge in the quality of search results due to
a) paid crap results
b) even the good results organizing pages so that any useful info is hidden far down in the bowels, behind click-bait headlines and complete crap leading paragraphs that read like they were written by distracted 1st-graders

 

[dioxide45]

Unless it copied another document word for word or used some portion that it becomes infringement, it isn't really copyright infringement. It wouldn't be any different than someone going to multiple sources to write up a letter, essay or other document.

 

[WaikikiFirst]

obtw, I own a bleep-load of GOOG stock. It is up > 4% today. Investors love this lawsuit. The bear story was that AI would "kill search", or more specifically, OpenAI would kill search for GOOG. Well, GOOG search is once again quite effective and the rest of the web knows it.
One mans plagiarism is another man's "hey that was fast and effective".

 

[jp10558]

Personally, unless you're a "substrate chauvinist", I have trouble distinguishing or drawing a line between a human reading a bunch of publicly available documents and writing an answer and AI doing the same. In many ways, from this POV, my 6th-12th grade self was breaking all sorts of laws and norms writing "research papers" in school. To my mind, this is similar to "human scrub dishes in sink OK, put in dishwasher BAD".

It's also interesting that it's OK for Google to have a huge database of compressed / hashed / indexed data for search, but AI - that's no good.

In terms of CoPilot in Word, my bigger concern is Microsoft slurping up data I did not publicly post. Same issue I've had with their (and other's) telemetry for a long time.

 

[rickandcindy23]

Agree on the CoPilot.

 

[HitchHiker71]

Unless it copied another document word for word or used some portion that it becomes infringement, it isn't really copyright infringement. It wouldn't be any different than someone going to multiple sources to write up a letter, essay or other document.

Exactly, all AI is doing is what a human does, simply on a mass scale via LLM, and therefore it's generally better than any one individual human. LLMs are literally simply doing the same thing that humans do - they were built by humans after all - and generally don't have privileged access to data - LLMs are literally scouring the web and then applying inbuilt logic to provide answers to your question(s). The benefit of AI is that it doesn't forget, doesn't have a bad day, doesn't make spelling or grammatical errors, and generally provides logical consistency when building a document upon request. I say generally, because there is such a thing as AI hallucination, where it seemingly knows what it's talking about, but can actually provide inaccurate guidance. AI isn't perfect, just like humans aren't perfect. Most folks don't realize that when they post on forums or upload content to various websites, that there are T&Cs that explicitly state that the content is searchable via public search engines, and/or that the content can be sold to third parties under certain conditions. You can opt out of these T&Cs, but by default, you opt-in as part of the user/usage agreement.

The fact that AI is generally "better" than most individual humans, isn't a bad thing, it's actually a good thing IMHO. It means we're creating more consistently accurate data/content that better protects humans in whatever they are trying to accomplish. A better written deed for example, that reduces the time spent, removes barriers to individuals getting something done without having to pay a third party to do so. Efficiency is good. Does that mean human labor, especially for knowledge workers, will become outdated? Over time, yes, and there's no stopping this fact.

 

[WaikikiFirst]

interesting that it's OK for Google to have a huge database of compressed / hashed / indexed data for search, but AI - that's no good

What is interesting is that you chose to ignore the fundamental difference, the fundamental extension that "AI Overview" introduced. That fundamental extension is almost certainly the heart of the lawsuit, and nobody in court will be able to ignore it, as you did.

 

[WaikikiFirst]

The only thing that plagiarizes faster than AI does is a bunch of Harvard humanities profs. Oh, the humanity!

 

[WaikikiFirst]

Exactly, all AI is doing is what a human does

Is that supposed to be some kind of legal argument.
"People don't plagiarize, so neither does AI."
Check your assumptions. and btw, the word "plagiarize" came from Cindy. I'd bet the lawyers will tend to use terms that fit the economic, commercial terms of the world of websites. On a slight tangent, the term "scrape" was used in the short pieces I read, and it will likely be used in court. The term "scrape" has a long history of use in the computer world, in the field of info being found, prepared, presented on a display.

 

[Brett]

I still think it's unethical for AI to search out documents for whatever subject and fill in the information.

AI is scary.

Make those AI robots pay up !!

https://apnews.com/article/anthropi...ent-training-f294266bc79a16ec90d2ddccdf435164

 

[HitchHiker71]

Is that supposed to be some kind of legal argument.
"People don't plagiarize, so neither does AI."
Check your assumptions. and btw, the word "plagiarize" came from Cindy. I'd bet the lawyers will tend to use terms that fit the economic, commercial terms of the world of websites. On a slight tangent, the term "scrape" was used in the short pieces I read, and it will likely be used in court. The term "scrape" has a long history of use in the computer world, in the field of info being found, prepared, presented on a display.

So let's look at the definition of plagiarize via post #12. With that in mind, I'm fairly certain that the application of plagiarism is a bit sensationalist. I never meant AI was plagiarizing, I was merely pointing out that LLMs are basically using human logic on a mass scale with access to a much larger and richer dataset "in mind" all at the same time - this is why AI uses up huge amounts of compute power (and literally electric power). Every data point is in memory when AI is processing the request in scope. No human has the ability to access every data point in their own mind at a single point in time, let alone the datapoints of millions of other humans that has been shared on the internet. That's what LLMs do in comparison. Screen scraping, which is the term you're referring to, really isn't applicable specific to LLMs, though LLMs can source data from other systems that may or may not employ screen scraping: https://x.com/i/grok/share/7afx4JP19no0MopNoONkXQyq9

 

[jp10558]

What is interesting is that you chose to ignore the fundamental difference, the fundamental extension that "AI Overview" introduced. That fundamental extension is almost certainly the heart of the lawsuit, and nobody in court will be able to ignore it, as you did.

While I realize that courts rarely operate that close to reality or logic, it does seem to me that copyright is about the copying part and as I recall from all the piracy wars circa 2001 or so ... the technical details didn't matter very much, it was the end display. But AI Overview doesn't display the copyrighted material in 99.99999% of the cases. And the other part - a computer "reading the data" already has had a longstanding "sort of exception" because to even run a program a computer has to "read the data". Again, I'm sure there's a case here, and IDK how it'll play out, but it strengthens my own personal dislike of IP law entirely because it leads to many frankly insane discussions and legal cases.

EDIT: Also, the OP topic is "Ethical Questions" so my argument is not a legal one to begin with, but was philosophical.

 

[jp10558]

Is that supposed to be some kind of legal argument.
"People don't plagiarize, so neither does AI."
Check your assumptions. and btw, the word "plagiarize" came from Cindy. I'd bet the lawyers will tend to use terms that fit the economic, commercial terms of the world of websites. On a slight tangent, the term "scrape" was used in the short pieces I read, and it will likely be used in court. The term "scrape" has a long history of use in the computer world, in the field of info being found, prepared, presented on a display.

Ok, people reading several public documents and providing their employer a summary generally have yet to have copyright cases brought against them that I'm aware of. "A computer did it" really should not be different under the law IMO - all this "stuff we understood, but now a computer did it so it's special and needs new laws" is quite crazy to me.

 

[WaikikiFirst]

people reading several public documents and providing their employer a summary

Such people do not have a commercial relationship with the producer of the documents, other than maybe having bought 1 or 2 copies. Copyright, a commercial concept, is a better path for discussion than is "plagiarism", which is more of an academic concept. I wonder if all the lawyers who have specialized in "Digital Law" for the last 30 years think it is crazy.
Never underestimate how the digital world differs from the physical world, nor whether lawyers will get involved.

 

[WaikikiFirst]

AI Overview doesn't display the copyrighted material in 99.99999% of the cases

You KNOW this? Really? I'm shocked. I'd love to see your research.