Xitter warns of a series of sexually explicit content manipulations on "Grok" and threatens legal action, including permanent account suspension
3d 21h ago by lemmy.world/u/lechekaflan in fuck_ai
This in reaction to outrage over what amounted to massive instant free-for-all AI image manipulation without even consent.
While the backlash against Grok manipulation had a strong showing, still... remember the adage: "Some are more equal than others."
It's not our responsibility to control grok, it's your responsibility not to prompt it in some badly defined way. smh.
Do they expect people are going to just not abuse the thing they are bringing attention to?
Obviously they know they can't keep AI from doing stuff, is this their attempt to avoid legal responsibility? Maybe they should ask grok how to keep grok from making child porn.
I'm sure they have.
Why do I feel like Grok's transgressions were by design? This coverup attempt sounds like what happens when a kid gets caught doing bad things, Adolf Musk wanted this shit to happen and he's upset that he got caught.
Invade them. Drag the fucker off in chains and parade him to the press. Oh, no wait he is politically protected. Nice society they have over there where they pardon criminals and don't prosecute fraudsters and pedos because of political connections. Must make people proud. Greatest democracy on Earth. Very Christian of them. Punish the poor, reward immorality, just like it said in the bible? Fucking weirdos.
I'm curious how much prop section 230 could save them from in this instance since they don't regulate or moderate the tool they built in any way and Grok seemingly may be able to post these edited images to xitter even when the user account is banned?
Like. They don't get protection from their own chatbot, do they?
Ok, i don't have an answer but that would be an interesting court case
as if they upload illegal content
Is it just me or is this fucked up English? Both Musk and “Safety” use the shit wording?
Elon got worried when he was told Twitter was losing its legal protection in India and is basically throwing the blame at users for him making and even controlling his fake AI creation.
The worst part is that it will probably work but I dont know why they let him basically blame the dog for shitting on the floor when he is still currently doing it.
Ah, yes. Shedding of the responsibilities. You gave the “kid” said “weapon”. You, X, should have known better. It is YOU, x, who is the party responsible. Releasing a faulty tool without proper safeguards…
By definition, Grok can't make child sexual abuse material, because it's fake. Those posts need community notes.
Post needs accessibility.
Images of text break much that text alternatives do not. Losses due to image of text lacking alternative such as link:
- usability
- we can't quote the text without pointless bullshit like retyping it or OCR
- text search is unavailable
- the system can't
- reflow text to varied screen sizes
- vary presentation (size, contrast)
- vary modality (audio, braille)
- accessibility
- lacks semantic structure (tags for titles, heading levels, sections, paragraphs, lists, emphasis, code, links, accessibility features, etc)
- some users can't read the image due to lack of alt text (markdown image description)
- users can't adapt the text for dyslexia or vision impairments
- systems can't read the text to them or send it to braille devices
- web connectivity
- we have to do failure-prone bullshit to find the original source
- we can't explore wider context of the original message
- authenticity: we don't know the image hasn't been tampered
- searchability: the "text" isn't indexable by search engine in a meaningful way
- fault tolerance: no text fallback if
- image breaks
- image host is geoblocked due to insane regulations.
Contrary to age & humble appearance, text is an advanced technology that provides all these capabilities absent from images.
Weird thing to go to bat for but the emphasis is on material. Even if it’s generated or drawn if it depicts CSA acts it can constitute CSAM and is even treated as such in a lot of jurisdictions. Though generated more so than drawn due its appearance of being genuine. It feels like saying “this isn’t porn of Y celebrity because it’s fake”, but it’s still porn.