Email or username:

Password:

Forgot your password?
Susan Kaye Quinn 🌱(she/her)

Back in Feb, when Grammarly said they were going to think about incorporating AI, I immediately canceled my account and uninstalled. And now they've actually done it.

I'm sharing because with everything going on, writers might not realize this is happening.

#AI #writing

54 comments
Lyssa Chiavari

@susankayequinn Haven't had a new book out in a couple years because health, but I always would run Grammarly as my final passthrough for edits, so really glad you posted this because I was going to do it next month 😬

Susan Kaye Quinn 🌱(she/her)

@lyssachiavari Same. It was good at catching typos and some comma/hyphen issues. I'm still using WordRake, which will catch some of that stuff, because I haven't heard of them incorporating AI yet.

Petra van Cronenburg

@susankayequinn Do you know a good alternative? I need it for working in a different language.

Susan Kaye Quinn 🌱(she/her)

@NatureMC I use WordRake to catch some typographical/grammatical stuff, and I haven't heard of them incorporating AI. I'm not sure if that will serve your purposes or not, but I do like them.

Petra van Cronenburg

@susankayequinn Thank you for the tip! Unfortunately, I can't afford software that costs money monthly. If I could, I would pay an editor.πŸ˜‰

Susan Kaye Quinn 🌱(she/her)

@NatureMC Human editors are much more expensive, and do a much wider range of tasks. I'd say WordRake is only really good for catching some grammatical errors (and not even all of those). I'm not sure what you're using it for (and how often) but all editing takes either time/money and usually both.

Rafael Caricio

@susankayequinn @NatureMC Maybe you can check out LanguageTool which is open-source and you can selfhost as well:

https://github.com/languagetool-org/languagetool

Sojourn :coffefiedyellow:

@susankayequinn @NatureMC wonder if you paid a human editor now, if they'd use an AI to assist them. You can ultimately view LLMs at tools to augment humans most of the time.

Petra van Cronenburg

@susankayequinn I know all this, I'm an insider in that market.πŸ˜‰ I use grammarly *before* I give my texts to anyone. Call it perfectionism.😎 If Wordrake can do less then grammarly, it is definitively too expensive. (Buying software for a fix price one time would be ok, but not subscriptions.)

Alessio Maffeis

@susankayequinn sending everything you type to a third-party service over the Internet is the root problem, and that’s how Grammarly has always worked, long before the AI trend. This is just a direct consequence.

vladcampos

@susankayequinn I've been happy LanguageTool user for a long time, and I suggest you take a look at their Privacy Policy.
πŸ‘‡
languagetool.org/legal/privacy

Kevin Lloyd :mstdnca:

@susankayequinn Training to make it better is one thing in an effort to make helping everyone else and their content better. They need to train it on something. If they're selling that content, I would take issue with it.

Unless I'm missing something here?

Noxy 🐾

@susankayequinn can you provide better evidence than a screenshot of a tweet?

L.J., un-author-ized writer

@noxypaws their terms grammarly.com/terms#user-conte at least seem to leave it open.

"You grant us a license to your User Content for the limited purposes of: ...

"Developing new products or features (for example, creating our tone detector)"

Together with the fact that they bill themselves as an AI writing assistant service, it's not hard to put 2 and 2 together and no one should be surprised if this was going on awhile.

@susankayequinn

Noxy 🐾

@ljwrites @susankayequinn I wouldn't put it past them, or anybody, at the moment. but it should be much bigger news if there was any sort of revelation recently.

L.J., un-author-ized writer

@noxypaws "In some cases, we may store some text to help us improve the product, though we do not store all text transmitted through Grammarly.”

Collins noted that the company removes user account data and personally identifiable information from *text it uses to train its natural-language processing systems.* (asterisks mine for emphasis)

Source: protocol.com/newsletters/proto

Revelatory enough for you? Like I said, this isn't new--the linked article is from February 2022. If you think this would have caused a stir, you underestimate how normalized this kind of thing has become.

@susankayequinn

@noxypaws "In some cases, we may store some text to help us improve the product, though we do not store all text transmitted through Grammarly.”

Collins noted that the company removes user account data and personally identifiable information from *text it uses to train its natural-language processing systems.* (asterisks mine for emphasis)

Chumchum Tumtum

@susankayequinn I always assumed that was what they were doing, so I never signed up

F00F/Eris :lowResBlobCat:☯️

@susankayequinn honestly, compared to all of the privacy issues grammarly has had since its inception, basically being a browser keylogger, AI is the least of my concerns

Andrew

@susankayequinn I feel vindicated. I always thought grammarly was just a keylogger with autocomplete.

Christie Dudley

@susankayequinn Copyright is only one of the problems. All sorts of non-public information is likely to be exposed to the public...

Winchell Chung βš›πŸš€

@susankayequinn
The trouble is, even if you cancel your account, there is no guarantee the company has not already made a covert copy of your writings.

Some Knucklehead

@susankayequinn that’s… very concerning. I know of a lot of academics that use it regularly.

Generic Sadboy 1916

@susankayequinn This comes as no surprise but with some schadenfreude; Given that multiple managers had recommended Grammerly to 'cure' my dyslexia.

Kevin Karhan :verified:

@susankayequinn I never used #Grammarly because it's #ToS were not only.inacceptable amd a gross violation of #BDSG & #GDPR, they never even supported #Ubuntu #Linux...

Bad RS

@kkarhan @susankayequinn grammarly has always been a data harvesting scheme, what else could it be? You give them access to everything you write and in return they correct your comma usage. Honey is another one, you let them track you across the internet and in exchange they occasionally save you 15% on toilet paper.

Dr. Eric J. Fielding, PhD

@susankayequinn My employer has banned Grammerly from our work computers, even before the new AI. They have always been sending your text back to their servers and we have to be careful with export control.

ocdtrekkie

@EricFielding @susankayequinn Yeah I have all access to Grammarly domains blocked in our corporate firewall. For anyone with legal compliance obligations, it's always been entirely unacceptable.

HistoPol (#HP)

@susankayequinn

Thanks.

Will do likewise.

....

But how about MS Word 365...

Susan Kaye Quinn 🌱(she/her)

@HistoPol I've never used 365 but when Word pushed their gen-AI "Editor" even into my desktop 2019 version, I started looking for other word processors. Which is tough as a writer because Word is the standard and comment handling is difficult in other software. But I'm making that transition slowly

Meow :verified:

@susankayequinn Corporations are going to be the ruin of us. I swear.

lucas

@susankayequinn One of the staff at work were complaining that word was slow when she was WFH. It indeed was - she could type "hello" and give a thumbs up to the webcam before the l's even appeared on screen.

After much stuffing about not improving anything, we eventually disabled grammarly's office plugin, and it was back to normal.

I'm guessing there was some latency uploading her document to "zee cloud" repeatedly.

Karl Emil Nikka

@susankayequinn Have you found a link to the announcement Cat Rambo is referring to? (I asked her on Twitter as well.)

ocdtrekkie

@susankayequinn Grammarly is, and always was, a keylogger. The fact that it pipes everything you type to a cloud service with an unquestionably broad terms of service was already a deeply risky thing to do. Very unsurprised at this outcome.

Holger Fiallo

@susankayequinn Grammarly was not accessible to those of us who are blind. Many people complain but nothing was done. Sad.

Fardels Bear

@susankayequinn

I've never used them and hate their ads on youtube, so I'm willing to believe they're capable of any evil.

Veronica Olsen πŸ³οΈβ€πŸŒˆπŸ‡³πŸ‡΄πŸŒ»

@susankayequinn Grammarly has always looked dodgy to me, so I never used it. I've looked a bit into LanguageTool instead. In particular, running it locally.

Etua

@susankayequinn What are more privacy-respecting alternatives? Of course I don't mean trivial spelling correction.

Susan Kaye Quinn 🌱(she/her)

@etua_en I don't have good recs honestly β€” human copyeditors, probably.

🌱 Ligniform :donor:​

@susankayequinn I've always felt weird about grammarly. Something about them owning the rights to what you write using it (I think?)

Alastair Cooper

@susankayequinn@wandering.shop This is what GitHub have been doing for a while now but a key difference is that GitHub has become so deeply embedded in the developer community its extremely difficult to completely out out.

kolya

@susankayequinn Is there any more reliable info on this than a screenshot of a tweet from a person who calls themselves Cat Rambo?
Not doubting, I'd just like to have a link to that announcement. @catrambo

Susan Kaye Quinn 🌱(she/her)

@kolya @catrambo Cat is an ex president of the Science Fiction and Fantasy Writers Association

kolya

@susankayequinn @catrambo
That's cool. Do you have a link to that announcement? I can't seem to find it.

Trajecient

@kolya @susankayequinn @catrambo Not an announcement, but I did find a statement from a Grammarly representative quoted in this article by a tech media company: protocol.com/enterprise/gramma

It isn't recent - from back in 2022. Not familiar enough with Protocol to make comment on its reliability.

The Terms of Service and Privacy Policy leave this open to occur (as part of use of data to provide and improve services) but do not specify if user data becomes AI training data or is just processed by AI.

@kolya @susankayequinn @catrambo Not an announcement, but I did find a statement from a Grammarly representative quoted in this article by a tech media company: protocol.com/enterprise/gramma

It isn't recent - from back in 2022. Not familiar enough with Protocol to make comment on its reliability.

Seph Harrisonβ™Š βœ…
@susankayequinn If you need a service like that, look at Zoho, specifically Zoho writer and it's Zia feature
h2lift_SciFi

@susankayequinn
I always assumed that Grammarly was scraping data. Now it's called 'feeding the AI'
Before that, it was probably for fun.
Anything in the cloud can be captured.
Sorry if this is new information.

Ame the Squirrel :dns:

@rifkafox been looking for the sauce on this for a while now and coming up empty paws. Do you know of where this announcement is?

RifkaFox

@onlyhavecans grammarly.com/blog/how-grammar

Everything in Grammarly trains the AI.

Which sucks because I enjoyed their service.

Go Up