Love getting flagged by an AI service for using AI to plagiarise coursework.

[tinmouth@infosec.exchange]

@SecurityWriter this sounds thorougly excellent. I love the future!

[jtig@infosec.exchange]

@SecurityWriter betting 5 pounds that you’re getting a lower grade for the humblebrag.

[mikalai@privacysafe.social]

@SecurityWriter
US court says that AI isn't copyrightable.
Do they interprete this as "you loose copyright, when AI scrapes your stuff"?

[geospaz@mastodon.social]

@mikalai @SecurityWriter

[suneauken@mastodon.world]

@SecurityWriter

Insufferable! Go get 'em, tiger.

[mnf@mastodon.social]

@SecurityWriter "Currently arguing with their submissions team" - are you sure? 🫣

[sometimeslovely@mas.to]

@SecurityWriter
Once had this happen irl. (20+ years) I proved I was my own source, but was still censured for not referencing myself correctly. I am still unsure as to how I was denying myself credit for work i had not yet published. The style was familiar to the board, so they assumed plagiarism, and needed to win!

[hummingbirdsecurity@mastodon.social]

@SecurityWriter Yikes.

[tom@subdued.social]

@SecurityWriter
So an AI solution to a problem created by AI creates another problem. That checks out.

[mainframed767@infosec.exchange]

@SecurityWriter this cyberpunk future sucks, at least in the books they had flying cars and O'Neil cylinders

[cr0w@infosec.exchange]

@SecurityWriter Good luck but I was told that even if you are referencing yourself, it needs to be cited. :-/

[kate@mk.absturztau.be]

@SecurityWriter@infosec.exchange ...when the gangsters argue with legal conditions you truly are among the AI gangsters lemon squeeze machines. I wish you luck!

[arem@mstdn.ca]

@mikalai @SecurityWriter AI output is uncopyrightable, meaning AI receives no copyright protection.

The original author retains their copyright, which output scraped then duplicated by AI.

[lyrenhex@social.lyrenhex.com]

@SometimesLovely @SecurityWriter afaik, it's not necessarily a matter of "denying credit", but allowing for peer review.

In theory, all statements are either:

1. trivially verifiable by virtue of being assumed knowledge;
2. novel contributions which are fully justified in the produced work itself; or
3. prior art, regardless of who it was that produced the art (including yourself).

In the first two cases, no references are required because they wouldn't make sense. In the last case, however, references are always required because otherwise how do you know where that statement came from? How can you assess its veracity without a trail to follow?

Otherwise, one can make novel claims and not be questioned on it because "it might be from their prior work".

[chemicaleyeguy@mstdn.science]

@SecurityWriter #AI is #clankers 🤖 all the way down.

[pogowasright@infosec.exchange]

@cR0w Are you serious? Who told you that? @SecurityWriter

[cr0w@infosec.exchange]

@PogoWasRight @SecurityWriter My Masters program. It wasn't a good program but if they were pushing that, then I can see why others would think the same thing.

[devlogic@masto.hackers.town]

@cR0w @PogoWasRight @SecurityWriter my spouse who is working on a PhD right now has been instructed to self-cite when referencing from their MLitt dissertation.

[dacmot@sunny.garden]

Isn't this just a clear net win for humanity @SecurityWriter? /s

[mirabilos@toot.mirbsd.org]

@SecurityWriter @mikalai the output is not copyrightable as a new work (because it’s just a machine transformation of existing works), but it can still be under copyright of the "training data" if close enough