> Grok AI generated large amounts of CSAM and nonconsensual intimate imagery
Grok Imagine has been considerably locked down in terms of intimate imagery over the last few weeks. E.g. Harley Quinn used to be one of the easiest characters to manipulate, with or without any resemblance to Margot Robbie. No more. X still serves up explicit hardcore, and Imagine used to get at least in that neighborhood, but that has been squelched. For prurient purposes, nerfed. Not at all limited to CSAM or real people. The pressure they're getting from all over seems to explain it.
But they did resist locking it down, recall Musk making fun of concerns? They clearly don't take governance seriously, its whatever Musk is gravitating to in his filter bubble.
Our response[^1] to X’s petition debunks many claims the company uses in its arguments. For example, there’s little evidence the order placed an undue financial burden on X. In our letter, we note that the compliance cost is merely “a rounding error against the $200 billion valuation of X Corp. following the xAI merger.”
The letter is more interesting than the cover, undersigned by Center for Digital Democracy, Check My Ads Institute, Constitutional Alliance, Consumer Action, Consumer Federation of America, Consumers for Auto Reliability and Safety, Demand Progress Education Fund (“DPEF”), Electronic Frontier Foundation (“EFF”), Electronic Privacy Information Center (“EPIC”), National Consumers League (“NCL”), Oregon Consumer Justice, Oregon Consumer League, Public Citizen, Travelers United and Virginia Citizens Consumer Council, and drafted by DPEF's Special Advisor Kate Oh (kate@demandprogress.org),
EFF's Senior Staff Technologist William Budington (bill@eff.org), EPIC's Senior Counsel Sara Geoghegan (geoghegan@epic.org), and NCL's Senior Public Policy Manager Eden Iscil (edeni@nclnet.org).
The EFF does not blindly take the stance that "anything should be allowed as long as you do it with a computer". Their input here is very reasonable, and in standing with their principles.
It's basically a rant from someone at the EFF that didn't like DOGE and doesn't like Elon. All AI models can generate bikini photos of people that look under 18 (the cause of the 'X CSAM scandal' alluded to here) if manipuluted to do so and yes being able to eliminate waste means letting people access systems.
It's a shame to see the EFF stray so far from their original goals of online rights advocacy, particularly when the UK police are harassing people for 'legal but harmful' communication (the EFF operates globally and should be active on this).
> It's basically a rant from someone at the EFF that didn't like DOGE and doesn't like Elon.
Looks like 15 organizations with one letter, per the letter, not “someone at EFF.” Likely written by committee, as that’s how these things go. Pretty big and respected organizations too.
> All AI models can generate bikini photos of people that look under 18 (the cause of the 'X CSAM scandal' allded to here) if manipuluted to do so…
Well, you’re severely underplaying this to the best of my understanding. Sounds like it was girls substantially, substantially below 18, and from what I recall for Grok this was a simple prompt away, which isn’t quite properly characterized by “manipulated to do so” verbiage.
Sounds like you’re carrying water for Elon and DOGE and those groups. Would encourage you to do some more reading on the subjects and perhaps reevaluate the subjects of your admiration. They truly do not care about you in the slightest.
> Sounds like it was girls substantially, substantially below 18, and from what I recall for Grok this was a simple prompt away
"Sounds like" is vague as fuck. You don't don't seem to know the cause of the ussue, so let's be specific: Grok and every other generative AI was producing bikini pictures of children if prompted to do so.
Bikinis weren't counted as nudity - which may seem logically correct from a programmer PoV but isn't the right approach. Essentially the rule should be "don't manipulate photos of anyone to be wearing less revealing clothing than the source material" but that isn't obvious. And yes, putting people in bikinis was a prompt away on every other AI service at the time, and is now locked down on Grok.
This has not helped the case you are trying to make.
>> Sounds like it was girls substantially, substantially below 18, and from what I recall for Grok this was a simple prompt away
> "Sounds like" is vague as fuck. Grok and every other AI was producing bikini pictures of children.
> Bikinis weren't counted as nudity - which may seem logically correct from a programmer PoV but isn't the right approach. Essentially the rule should be "don't manipulate photos of anyone to be wearing less revealing clothing than the source material" but that isn't obvious.
If you actually try reading the EFF's letter, you will immediately see that the issue was not just "putting people in bikini's"
Grok Imagine has been considerably locked down in terms of intimate imagery over the last few weeks. E.g. Harley Quinn used to be one of the easiest characters to manipulate, with or without any resemblance to Margot Robbie. No more. X still serves up explicit hardcore, and Imagine used to get at least in that neighborhood, but that has been squelched. For prurient purposes, nerfed. Not at all limited to CSAM or real people. The pressure they're getting from all over seems to explain it.
The EFF featured update / press release at https://www.eff.org/deeplinks/2026/06/eff-and-allies-xs-ftc-... links to the letter, with color:
Our response[^1] to X’s petition debunks many claims the company uses in its arguments. For example, there’s little evidence the order placed an undue financial burden on X. In our letter, we note that the compliance cost is merely “a rounding error against the $200 billion valuation of X Corp. following the xAI merger.”
[^1]: public interest advocates opposing x petition 2026: https://www.eff.org/files/2026/07/02/public_interest_advocat...
The letter is more interesting than the cover, undersigned by Center for Digital Democracy, Check My Ads Institute, Constitutional Alliance, Consumer Action, Consumer Federation of America, Consumers for Auto Reliability and Safety, Demand Progress Education Fund (“DPEF”), Electronic Frontier Foundation (“EFF”), Electronic Privacy Information Center (“EPIC”), National Consumers League (“NCL”), Oregon Consumer Justice, Oregon Consumer League, Public Citizen, Travelers United and Virginia Citizens Consumer Council, and drafted by DPEF's Special Advisor Kate Oh (kate@demandprogress.org), EFF's Senior Staff Technologist William Budington (bill@eff.org), EPIC's Senior Counsel Sara Geoghegan (geoghegan@epic.org), and NCL's Senior Public Policy Manager Eden Iscil (edeni@nclnet.org).
They're arguing X is a massive privacy risk and should not get any exemptions.
Basic human decency?
[1] https://arstechnica.com/tech-policy/2026/06/spacex-may-donat...
It's a shame to see the EFF stray so far from their original goals of online rights advocacy, particularly when the UK police are harassing people for 'legal but harmful' communication (the EFF operates globally and should be active on this).
Some relevant previous discussion: https://news.ycombinator.com/item?id=47707802
Looks like 15 organizations with one letter, per the letter, not “someone at EFF.” Likely written by committee, as that’s how these things go. Pretty big and respected organizations too.
> All AI models can generate bikini photos of people that look under 18 (the cause of the 'X CSAM scandal' allded to here) if manipuluted to do so…
Well, you’re severely underplaying this to the best of my understanding. Sounds like it was girls substantially, substantially below 18, and from what I recall for Grok this was a simple prompt away, which isn’t quite properly characterized by “manipulated to do so” verbiage.
Sounds like you’re carrying water for Elon and DOGE and those groups. Would encourage you to do some more reading on the subjects and perhaps reevaluate the subjects of your admiration. They truly do not care about you in the slightest.
No. The institutions are:
Center for Digital Democracy
Check My Ads Institute
Constitutional Alliance
Consumer Action
Consumer Federation of America
Consumers for Auto Reliability and Safety
Demand Progress Education Fund (“DPEF”)
Electronic Frontier Foundation (“EFF”)
Electronic Privacy Information Center (“EPIC”)
National Consumers League (“NCL”)
Oregon Consumer Justice
Oregon Consumer League
Public Citizen
Travelers United
Virginia Citizens Consumer Council
> Sounds like it was girls substantially, substantially below 18, and from what I recall for Grok this was a simple prompt away
"Sounds like" is vague as fuck. You don't don't seem to know the cause of the ussue, so let's be specific: Grok and every other generative AI was producing bikini pictures of children if prompted to do so.
Bikinis weren't counted as nudity - which may seem logically correct from a programmer PoV but isn't the right approach. Essentially the rule should be "don't manipulate photos of anyone to be wearing less revealing clothing than the source material" but that isn't obvious. And yes, putting people in bikinis was a prompt away on every other AI service at the time, and is now locked down on Grok.
> No. The institutions are:
[list of big and respected organizations]
This has not helped the case you are trying to make.
>> Sounds like it was girls substantially, substantially below 18, and from what I recall for Grok this was a simple prompt away
> "Sounds like" is vague as fuck. Grok and every other AI was producing bikini pictures of children.
> Bikinis weren't counted as nudity - which may seem logically correct from a programmer PoV but isn't the right approach. Essentially the rule should be "don't manipulate photos of anyone to be wearing less revealing clothing than the source material" but that isn't obvious.
If you actually try reading the EFF's letter, you will immediately see that the issue was not just "putting people in bikini's"