They claimed there will be no CSAM because of the given reasons.

I wanted to highlight that those reasons do not actually prevent it.

My tone might be harsh (the sarcasm at the end definitely is) because this is a marketing push for their crypto platform. “marketing” - as in they will be making money from users, so it is in their interest to tell lies or ignorant half-truths, to make more users come over.

Any normal platform tackles this problem with proper moderation. Platforms that make money, often hire moderators.

I guess my question should be how they managed to do this without having to create an account and profile users

cookies

If it’s a cookie, the history should not be there if you clear your cookies or open it in a private tab.

The whole chat could be stored “in the browser” but more likely they have it on their server and associate it to you via the cookie. *edit: I guess it is this then according to what afk posted.

If it is not a cookie, there is also your IP and lots of alternative fingerprinting ways of uniquely identifying your browser (see creepjs ). You could use a VPN and disable js in your browser, but that breaks half the internet nowadays.

the protocol is text only, to embed media, you need to host it on the regular ( Centralized ) internet

except we already figured out how to encode images (or any file) as text when E-Mail was created. That is how images in E-Mails, attachment or embedded, are done. I can easily imagine a userJS script that will render them in the browser, but even if not you just copy the text and decode.

if a community is badly moderated, the user will never see it, it wont be recommended to him. the user can visit bad communities directly just like you can visit a bad website directly, but it’s not recommended to you so it’s safe to use.

Ah… so you’re guaranteed to have a dark CSAM subculture on there at some point.

being p2p, seedit is not private, so it can’t really be used for illegal activity

As if that has ever stopped anybody. See all the people that got caught for sharing it on the clearnet. Or on Signal, Telegram or similar, where you have to enter your phone number, which is personally tied to you.

All in all - Great way to adress the concerns, by admitting they are in fact possible. “Hurray crypto” or whatever.

I’m using a selfhosted forgejo but in case something goes wrong with that, everything is also mirrored to sr.ht (which has a shit GUI if you are liking github/lab).

to add to what Elvith wrote:

you can read the HTML like structures inside a PDF and then find out details about the elements you want to remove and then remove them by using that found common property.

This is very hard to do by hand. But if you are curious you can download https://file-examples.com/wp-content/storage/2017/10/file-sample_150kB.pdf

and open it with a text editor like kate. You will see a lot of encoded content data, but also the “html-like” structure in plaintext (in between the encoded stuff but also more at the bottom)

You might find that editing the PDF by hand will break it completely, that is because it is complicated. Iirc you’d need to fix the index, recalculate the checksum or do some other magic bullshit. But that is often taken care of by the library.

Here is a shitty python example for that demo pdf that redacts the image at the last page by drawing a white rectangle over it. There is no way in pymupdf to delete an image or a textblock … but this is just an example. Other libraries might be able to do it (the one I used a decade ago in java could). I just wanted to point you in the general direction, hope you can see from here how iterating over all the pages, picking the right element and redacting it would work.

import pymupdf  # PyMuPDF

# Open the PDF
doc = pymupdf.open("./file-sample_150kB.pdf")

# Get the last page
page = doc[-1]

# Get all images on the page
images = page.get_images(full=True)

if images:
    # Get the xref of the first image
    xref = images[0][0]

    # Find all instances of the image and redact their bounding boxes
    for info in page.get_image_info(xrefs=True):
        if info["xref"] == xref:
            rect = pymupdf.Rect(info["bbox"])
            page.add_redact_annot(rect, fill=(1, 1, 1))  # white fill

    page.apply_redactions()

# Save the modified PDF
doc.save("./modified.pdf")
doc.close()

A way simpler approach might be to crop all pages at the bottom.

import pymupdf  # PyMuPDF

doc = pymupdf.open("input.pdf")  # open the PDF

for page in doc:
    rect = page.rect  # original page size
    new_rect = pymupdf.Rect(rect.x0, rect.y0 + 100, rect.x1, rect.y1)  # crop bottom 100px
    page.set_cropbox(new_rect)

doc.save("output.pdf")  # save the cropped PDF
doc.close()

Here are the docs: https://pymupdf.readthedocs.io/en/latest/the-basics.html

A PDF is (or at least can be) similar to a HTML document on the inside. A long time ago we used that at my company to edit PDFs through java code.

Is it possible for you to share the document so we can take a closer look at it? Or if you don’t want it on the internet, is there a way to share it privately?

privacy - in my opinion that depends more on your behaviour and the installed extensions than the browser.

functionality - a little bit, but of course that matters only if you need any of the listed features (see link). The big one is the additional extensions/plugins that you can install.

https://github.com/fork-maintainers/iceraven-browser

Sorry, I am just gonne dump you some links from my bookmarks that were related and interesting to read, cause I am traveling and have to get up in a minute, but I’ve been interested in this topic for a while. All of the links discuss at least some usecases. For some reason microsoft is really into tiny models and made big breakthroughs there.

https://reddit.com/r/LocalLLaMA/comments/1cdrw7p/what_are_the_potential_uses_of_small_less_than_3b/

https://github.com/microsoft/BitNet

https://www.microsoft.com/en-us/research/blog/phi-2-the-surprising-power-of-small-language-models/

https://news.microsoft.com/source/features/ai/the-phi-3-small-language-models-with-big-potential/

https://techcommunity.microsoft.com/blog/aiplatformblog/introducing-phi-4-microsoft’s-newest-small-language-model-specializing-in-comple/4357090

if (link was posted this week) {don’t post}

It was posted 3x to the [email protected] community. Or at least it looks to me like 3 different accounts posted the same thing to this very community.

I don’t really care about how it works, I’m just tired of the chan-esque experience where I have to question my sanity because I see the same posts every day.

Just because people that don’t actually participate in a given community, thus not seeing the older posts, share the same article because they look for a community that fits and dump it there.

Some subreddits had bots that detected and removed reposts and guided OP to the original post for them to add their discussion points.

Yes, by different users at different times, thats my point.

https://lemmy.ml/post/31520326

https://lemy.lol/post/46499125

same shit for the third time

maybe drawio

Is there a dumb phone not made by a corporation?

Wow! I didn’t know

It does need a server though. Either the centralized official one you can selfhost one.

Matrix, Briar, SimpleX, Threema

a long runway that allows us to become profitable when needed

Switch to self-hosting headscale when they enshittify in an attempt to become profitable, duh

not op

coming soon section at the bottom of https://curi.ooo/

personally I love useful local ai integration