AI summaries of Sanctioned Suicide users on here - is this an issue for you?

[noname223]

From time to time I search for noname223 on the internet. I used this profile name on here because I knew there are millions of nonames out there with various numbers. And it has some personal meaning but I elaborated on that myriads of times. It is harder to connect the dots then. There are some screenshots of Stop Sanctioned Suicide on Facebook which screenshotted some posts of mine. They are harmless though. The group isn't that active anymore from what I know. I think in case I die the internet will most likely forget me pretty fast. Which isn't necessarily a bad thing.

Today, I searched for a user's name on here because I was interested what the search engine would show. I used the Brave browser and it gave me a lengthy AI summary of the posts and personal information about the user with some personal details. I just tried it again whether it still works. I had to add Sanctioned Suicide after the name. And it still works.

I tried it with my username and another user again. And no summary shows up. I think the crucial difference is the first user name was sort of unique. There are so many noname's on here. I think the search engine probably has issues to differentiate between them.

I think the big AI chatbots have guidelines to prevent such privacy infringements to happen. They can still be circumvented. But smaller AI bots seemingly not. I think for some it is also a unique selling point to infringe copyright and personal rights.

Does this bug you? What could we do about that? I think you need a lot of posts. Maybe more than 1000, probably to have an AI summary of you. But actually I only tried it with 3 people.

 

[Dejected 55]

It is what it is. Long long ago I just assumed anything I ever posted online would be accessible to anyone forever, even if I tried to delete it for some reason after the fact. There are layers to privacy, but if someone wants to find out about you they absolutely can. Takes some people longer than others... and you can scrub some sites better than others... but once it is there it is there, for better or worse. If you stay true to yourself, even if sometimes you embarrass yourself online, then it is what it is.

 

[Forever Sleep]

It is bothering me now. Since the last YouTube video, I'm realising how stupid I was to feel like this was a safe space to open up in. Not that I watched the video or intend to try to but, it feels like such an invasion of privacy to be 'studied' I suppose. Especially when the intentions behind it are to expose and shame people.

I think my input here more or less does represent what I believe although, my opinions and emotions can and do change. I'm sure I could be embarassed about things I've said here but, I wonder if that's the same for everyone.

It's definitely affected me though. I'm trying to make more of an effort to stay off the forum now. I feel like I've lost something that felt like my sanctuary though and I feel so sad about that. But then, that's just the world really. Nowhere is particulary safe or private. I imagine there's always some sort of exploitation risk.

It definitely disturbs me that AI will make all this spying on people far easier to do. What's perhaps worse is- I imagine it can be targetted to highlight a bias. So- give me all the examples where this person has come across as bigoted say.

AI will be lead by how we think about someone. It will back up whatever narrative we give it presumably. AI feels like the ultimate 'yes' man. I assume it will only give us a balanced argument/ representation if we ask it to.

I suppose that means we will constantly have to be checking what we write to ensure it paints us in the most acceptable light but then- what will that mean for us individually? Will we buckle under the pressure to comply to the most acceptable ideas or else run the risk of being exposed? Who decides what those 'acceptable' ideas are? How good is it for our mental health to not feel able to express our true selves anywhere?

I imagine the majority of people are too busy and too wrapped up in their own lives to ask AI to build character profiles of other people. Maybe they will with celebrities or, people they fancy. I imagine it will be a nightmare for victims of stalking.

I imagine a lot of people interested in it will start with themselves though. So- presumably if/ when AI gets it wrong with them- they will realise it isn't necessarily a totally accurate representation of whoever else they use it on.

It's not really AI itself that's scary though- to me. It's the people using it and their intentions. Yours are more playful- obviously. Like an experiment to test the technology itself. Even advertisers don't massively trouble me. I don't feel compelled to buy things I don't want. Even if they've targetted adverts they think I'll fall for.

For me- it's more that these invasions into privacy will maybe become normalised. Plus- everything else that AI can do. Infringe copyright. Mimic another person's art. Even a writing style or thought process is an art. Will journalists be happy if some Joe Bloggs starts expressing their own ideas in their style of reporting?

The most frightening though are those who intend to bully or harass people with it. I dread to think how it will affect some people.

 

[Sweet Tart]

Thank you for posting your experience with this. I don't use AI, & it never occurred to me that it could have access to this site. It's good to know about these things and be vigilant with privacy for as long as it's possible to do so.

 

[Raine Meadows]

Maybe a hot take but I have less than a 1000 posts and still showed up under Google's AI Mode, granted my username is unique. It's kinda comforting to know that if someone searched the right words, I could indirectly be remembered by someone possibly far beyond the time period I ever lived in.
I'm not the first of my kind and I won't be the last.

 

[barely_afloat]

That's honestly disturbing but I'm not surprised. We can have our gripes with our privacy being infringed upon, but I don't think it'll change much.

I think the best way to keep ourselves safe from someone with bad intent would be to never use our usernames on this site elsewhere, having generic usernames (I love yours, noname), and prevent using identifying information.

A bit scary to think that someone could potentially use AI to gather data on specific users and generate a profile with their personal details. I hope that doesn't happen.

Stay safe.

 

[Dejected 55]

A bit scary to think that someone could potentially use AI to gather data on specific users and generate a profile with their personal details. I hope that doesn't happen.

You don't even need AI. AI is the lazy way. All you have to do is know someone you want to mimic online and look at their profile or posts.

People have been doing this forever on dating sites... finding a person whose profile they like and then creating/editing their own profile to be as compatible as possible with that person before sending them a match/chat request. It's harder to be fake in person... but online where you have lots of time to think... people just have to see what you do online and then copy you. Using a fake name helps sometimes... except if you behave the same and people can figure out who you are from the clues you leave in the manner you talk online.

 

[barely_afloat]

You don't even need AI. AI is the lazy way. All you have to do is know someone you want to mimic online and look at their profile or posts.

I agree. AI makes it easier and accessible for people who might not put the effort in to collect information on others.

I'm mostly worried that someone can collect information on a user and then use it to extort them. I can't speak for others, but I know it would be scary for me if somebody contacted me, knew personal details about me, and threatened to reveal my identity to people in my real life. while they might not actually be able to do that, the initial scare and the idea that it may be true can trick some susceptible people.

 

[RosebyAnyName]

Information needs to be linked together for it to be useful. Generally, the most important part is to make sure your sasu account doesn't link up to any other online accounts with more personal info. It's why the recommendation to pick a non-identifying username is so important.

It's still scary that even basic info like country or age can be congregated by AI to profile an individual, although I can't help but be curious what the AI would say about my profile. Would it say positive or negative things about me?


Edit:
Ok so I just tried to look up info regarding my own account using the brave ai.

It didn't get my age right, just giving a vague age range that I don't even belong to (said I was between 18 and 25, but I'm older), even though I'm pretty sure I've mentioned my exact age before. I also asked stuff like: "does user believe x" for opinions I've stated on the forum before, and it said the results were inconclusive / it didn't know. A lot of the answers were hyperfocused on essentially just saying: "this user is on the controversial forum, but they have posted in the recovery section and seem interested in recovery." For context I post in the suicide section and the recovery section. It also said that I posted something that I 100% definitely did not post (I think I commented on someone else's post, and the AI mistook that the op's post was mine).

I wonder if I'm just bad at AI, Brave AI is too censored to protect the privacy of individuals, or I just don't have enough posts for it to say much about me. Honestly I was secretly hoping it would roast me.

 

[Dejected 55]

The AI can't intuit those leaps-of-faith that a human can... like if I know you semi-well and I see someone on another site talking/engaging with others in a familiar way I start to think it might be you... and neither site has to reveal any personal information about you to allow me to conclude you are Person A on one site and Rando B on another. Now, if you are a person I really know well, it gets a little easier... because I can search for things I know you to be interested in... and look for people in those communities who resemble your posting style. Most of us are fairly consistent in our manner of talking even when we use fake names... so the better you know someone, the easier it is to "unmask" them.

AI is good at sorting through the obvious stuff... but the less obvious needs the human touch, and I still say a stalker who wants to unmask you or wants to follow you around from site-to-site is more dangerous than the AI. A "good" stalker wouldn't rely on AI either... because he/she would know that stuff isn't reliable.

In the admittedly small sample size in this thread, the AI experiment that OP did wasn't super accurate at mimicking us all the time... whereas many of us could look at posts made by users on here and I suspect do a better job at a mock debate in each other's voices if we were so inclined.

All this is to say... I'm not super worried about AI sleuthing people out. It might get better at it... but even if it does... it's really the people who are trying to find you that are the problem... and they don't need AI to find you if they are so inclined.

 

[Scythe]

So uhh I looked up my account on the brave AI and kinda fucked up honestly.

It claimed I was advertising method info. Which is half true I have a few documentation on plant poisons but I did not say shit about SN. I also certainly was not advertising any methods. It also claimed I was on a suicide livestream and encouraged suicide. Which is very false because there's no SS suicide livestream to begin with.

 

[fuzzypeach]

yes it does bother me because AI is not accurate most of the time

 

[cme-dme]

I really hope that's not a real person because this really made me crack up hahahaha

I tried again with different wording and this one is probably roughly 40% correct :p

 

[fadedghost]

Oh no! Im on the Internet!

 

[Dejected 55]

AI gets so many things hilariously wrong. When I'm really bored I'll ask it obviously stupid questions and sometimes get amazingly stupid answers.

 

[Cauliflour]

View attachment 193584
I really hope that's not a real person because this really made me crack up hahahaha
View attachment 193585
I tried again with different wording and this one is probably roughly 40% correct :p

You got me curious so I went and asked ChatGPT but I guess I'm not famous enough :/

 

[cme-dme]

You got me curious so I went and asked ChatGPT but I guess I'm not famous enough :/
View attachment 193916

I had to roll it a couple times for say it knew who I was lol