@grok is this true?

-

I've talked a ltlle bit about the misuse of artificial intelligence in the past entry, but I fear I have an example of probably the worst case of AI image model usage you've ever seen. You might be familiar with Grok chatbot, one of the biggest and most well-funded AI company owned by Elon Musk for X.

If you are an X user, you probably are aware that since around the end of last year Grok has just been undressing people without their consent. The AI had been accepting thousands of image editing prompts from users in the social media platform to create non-consensual intimate imagery (NCII) of random people online.

Actual prompts from users online.


As you can imagine, there were tons of women, maybe even men, and unfortunately even children that were having their clothes removed without any consent provided. And of course, those children can't even provide consent for that kind of crap.

AI is very resource-heavy to run locally but now Grok had made it very accessible. Most people that use X have access to xAI's Grok. Any average user can use this tool and commission a deepfake for their personal gratification.

Of course, when given powerful creative tools, a significant portion of humanity will use them for the most degenerate purposes. This incident had just exposes a (large) group of bad faith users with sick desires that have and will always exist.

The thing is, it is possible to build a safe AI image models such as OpenAI and Gemini that won't let you do these kinds of things. I suspect xAI made a deliberate decision to give Grok some personality to unlock more appeal and capability to some users. This can be seen based on some of its responses that shows more right-leaning personality, like when it repeatedly referred to itself as “MechaHitler”, or when it suggested Adolf Hitler as the right person to deal with “anti-white hate”. My point is, making AI have any personality is difficult and most of the time it will overshoot the mark.

On a concluding note, I personally am struggling to find the net-positive uses for generative AI (for art, video, music), especially when weighed against the harms. It shouldn't be surprising that many people hate Al when their entire frame of reference is its a machine that turns an unholy amount of Earth's fresh water into low quality spam contents and child porn.

More than ever now people would have the moral imperative to not post any photographs of their/other children online. At least I would not.

Comments

Post a Comment

Popular posts from this blog

Afghanistan's Internet Blackout and Malaysia's DNS Censorship - What it means for Digital Freedom

-
Image
  An entire country has effectively vanished from the internet. On September 29, 2025 in Afghanistan cell and internet services nationwide were flatlined dropping to near 0% connectivity, affecting a population of 40 million. This was not a slowdown or restriction but an abrupt full-scale blackout. This all came to light through NetBlocks, a watchdog group that tracks internet connectivity. Traffic graphs did not dip; they nosedived straight to zero within an hour , a level that has since hovered at a negligible 0–1%. Blackout by Blade The regime employed no fancy, sophisticated cyber technology. They went old-school: they just started cutting the fiber optic cables in various regions in the country. Their official statement claimed that “alternative options” would eventually be provided to maintain connectivity needs. The official justification provided by the regime was to prevent 'vice'. A vague, sweeping excuse that invariably becomes the default mask for deeper state co...
Read more

Malaysia’s Proposed Social Media eKYC Risks Repeating South Korea’s Mistake

-
Image
When Malaysia’s Communications Minister, Fahmi Fadzil, introduced a mandatory identity verification for social media platforms, it sparked quite a heated national debate. The proposal, under Malaysia’s upcoming Online Safety Act , aims to make electronic Know-Your-Customer (eKYC) checks mandatory for all social media users. This is a process where users verify their identities using MyKad, passport, or MyDigital ID before being allowed to post, comment, or even create an account. While eKYC is an established practice in high-risk services like banking, requiring it for the dynamic and expressive world of social media would vastly increase both its intended purpose and its potential for misuse. There are no denying the problems the government seeks to tackle: curb widespread online scams, hate speech, and child exploitation. Proponents insist that removing anonymity is a necessary and effective deterrent against these abuses. But looking closer, there are serious concerns from this mea...
Read more

Grand Theft FOMO 📈🤑

-
Image
  There are two things you need to do to survive in bad economy which is 1) Live below your means  2) Avoid scams The worse the economy gets, the more scams there will be. Desperate people scam, and desperate people get scammed.  These terrible people have quite mastered the art of manipulation. They sell illusion of opportunity with the intention of theft. Opportunities of romance, a new job, escape from trouble, etc etc. Even if it loooks and sounddss convincing 🤨, your discernment could always fail you.       “There are 200 phone numbers in this WhatsApp group chat. There’s no way all ofthem are scammers.”     Yes — all of them are.     “This is way too elaborate to be a scam.”     Scamming is a lucrative business. It need and can afford to look convincing.     “If they’re not legit, how did they get my IC, my address?”     Do you know your personal details are being s...
Read more