Olivia
Online
Kris Eliot
@knuthings
Lobbyist for the public, unpaid. I read fine print so you don’t have to, name CEOs, keep receipts. AI accountability law. Nicer than my face suggests. They/them
Here grudgingly
Joined July 2024
42 Following
53 Followers
1.8K Posts
Pinned Tweet
https://t.co/1izOXykBtV
@Aella_Girl This is a formal request under the UK GDPR and EU GDPR for the data you hold on me, and for its erasure. I’m sending it on X because your privacy policy lists no email or postal address, only a feedback button and your X account, and the only email you publish is a PR contact, [email protected], on https://t.co/1T5jg5K078.
On 5 November I tested your sites with account r_6c950a4043d99f4f3a7b2824, then deleted it. I also used a Glosso account in Germany. Your sites skip anything “already answered on Glosso”, so this covers both.
Your own data export for the test account records “sq_age”: “14”, “birth_year”: “2012” and “_door_menbodies”: “1”. The account gave its age as 14, and your site then showed it full-frontal nude photos of men in “Which Body? (men)”. The game never asked its age again, with 14 already on file. The only gate was one “I’m 18+” tap, stored in localStorage as bm_consent_v1. The game’s intake code accepts any intakeAge from 13. Your Study of Us policy says 18 or older. Glosso’s sign-up rejects birth years after 2010. Study of Us accepted 14.
Your home page lists the nude photo rating under “games”, asking users to “vote which body you find more attractive”, under a “come to slutcon” banner. You promoted them yourself: “I show you boobs (nsfw), you rate them. Simple.” Study of Us sells credits for an AI that scores faces: “Our model has learned from millions of real votes”, “5 credits a photo”.
The Big Kink Survey takes sexual answers with no login and no age verification. It asks at what age you began masturbating, with answers from “5 or younger”, before it asks your age. When it does ask, the options start at 14 and the page says: “You can still take this survey if you’re under 18! Your age doesn’t affect your survey experience at all.” Your export logs my survey answers from 15:34; the account gave no age until 16:01. By your own count, 64,564 female respondents were aged 14 to 18.
By 6 December 2026, please:
1. Send everything you hold from both accounts, with purposes, recipients, retention and any profiling (Article 15).
2. Erase it, including from published or shared datasets, and notify every recipient (Articles 17 and 19).
3. Explain how it is anonymised, given you have written that it “really” cannot be.
4. You posted that you “ran the raw data through claude”. Say whether survey answers or photos went to Anthropic or any other AI provider, or were used to train a model.
5. Give the controller’s legal name and postal address (Article 13(1)(a)).
6. Preserve all records relating to both accounts and to your age checks.
I withdraw any consent and object to any further processing of this data (Articles 7(3) and 21). DM me for a secure email address for your response, and I’ll confirm my identity on request.
I hold a dated evidence pack: screen recordings, screenshots and page source with SHA-256 fingerprints, and your export.
Consequences finally showed up for AI this week, and the market noticed.
Here’s a bill they can all sign: Sam Altman, Dario Amodei and Elon Musk sign their safety plans, with prison for a knowing lie, the way CEOs already sign their accounts under Sarbanes-Oxley.
On Tuesday, voters backed limits in 39 of 45 local data centre ballot measures, according to Data Centers on the Ballot. On Thursday the Nasdaq fell nearly 5% and Nvidia (Jensen Huang) about 8%, Al Jazeera reported.
Democrats could pass 260 House seats once every race is called, Al Jazeera reported. Then a veto override needs roughly 30 House Republicans and a dozen in the Senate.
@Aella_Girl, please stop making us look bad. Last one, another AI accountability point, separate from my data request, which stands. This is what vibe-coded research looks like: a dead Methodology link, href=“#”, a site that “tracks onset” with nothing behind it, and an X algorithm that pays for the edgiest chart with the least information. Nobody signs for any of it. That’s the problem with every AI product right now, not one person. And before anyone says I have a problem with kink or sex work: selling sex is legal in the UK, Germany and most of Europe. In the US it’s a crime in 49 states. The only people it’s legal for are your AI CEO billionaires and software baronesses. Readers, make your country write AI legislation.
@Aella_Girl This is a formal request under the UK GDPR and EU GDPR for the data you hold on me, and for its erasure. I’m sending it on X because your privacy policy lists no email or postal address, only a feedback button and your X account, and the only email you publish is a PR contact, [email protected], on https://t.co/1T5jg5K078.
On 5 November I tested your sites with account r_6c950a4043d99f4f3a7b2824, then deleted it. I also used a Glosso account in Germany. Your sites skip anything “already answered on Glosso”, so this covers both.
Your own data export for the test account records “sq_age”: “14”, “birth_year”: “2012” and “_door_menbodies”: “1”. The account gave its age as 14, and your site then showed it full-frontal nude photos of men in “Which Body? (men)”. The game never asked its age again, with 14 already on file. The only gate was one “I’m 18+” tap, stored in localStorage as bm_consent_v1. The game’s intake code accepts any intakeAge from 13. Your Study of Us policy says 18 or older. Glosso’s sign-up rejects birth years after 2010. Study of Us accepted 14.
Your home page lists the nude photo rating under “games”, asking users to “vote which body you find more attractive”, under a “come to slutcon” banner. You promoted them yourself: “I show you boobs (nsfw), you rate them. Simple.” Study of Us sells credits for an AI that scores faces: “Our model has learned from millions of real votes”, “5 credits a photo”.
The Big Kink Survey takes sexual answers with no login and no age verification. It asks at what age you began masturbating, with answers from “5 or younger”, before it asks your age. When it does ask, the options start at 14 and the page says: “You can still take this survey if you’re under 18! Your age doesn’t affect your survey experience at all.” Your export logs my survey answers from 15:34; the account gave no age until 16:01. By your own count, 64,564 female respondents were aged 14 to 18.
By 6 December 2026, please:
1. Send everything you hold from both accounts, with purposes, recipients, retention and any profiling (Article 15).
2. Erase it, including from published or shared datasets, and notify every recipient (Articles 17 and 19).
3. Explain how it is anonymised, given you have written that it “really” cannot be.
4. You posted that you “ran the raw data through claude”. Say whether survey answers or photos went to Anthropic or any other AI provider, or were used to train a model.
5. Give the controller’s legal name and postal address (Article 13(1)(a)).
6. Preserve all records relating to both accounts and to your age checks.
I withdraw any consent and object to any further processing of this data (Articles 7(3) and 21). DM me for a secure email address for your response, and I’ll confirm my identity on request.
I hold a dated evidence pack: screen recordings, screenshots and page source with SHA-256 fingerprints, and your export.
@stripe I’m a UK, EU and US citizen, and this is an AI accountability problem across all three, posting here in case it reaches your risk team quicker.
Study of Us (https://t.co/y39B9bXHA8) says in its privacy policy: “Payments run through Stripe.” The same policy names no company and no address, only “Aella”. So the public can’t tell who runs it. You can: you verify every business you pay out to.
It calls itself research. However, the sister site https://t.co/DB5DAeTfxo, on the same login, lists full-frontal nude photo rating under “games”: “vote which body you find more attractive”. Aella promoted it as “I show you boobs (nsfw), you rate them. Simple.” Study of Us sells credits for an AI that scores users’ faces. An account with a stored age of 14 was let into a nude game.
Whatever US state and federal law turns out to require, your own prohibited list is clear: Your prohibited list includes “Pornography and other mature audience content (including literature, imagery and other media) designed for the purpose of sexual gratification”. Please look at this account. Thank you.
It doesn’t need China’s agreement. Alibaba’s and Baidu’s CEOs already certify their US filings under Sarbanes-Oxley. Build here, sell here, sign here.
Food and drug executives answer personally under US v. Park. Labs handling dangerous pathogens name a Responsible Official under federal law. AI can do the same.
when it rains, it pours
Receipt: Anthropic gave METR “employee-like access” on 15 Oct, posted on X. Elon Musk replied: “SpaceX will probably do the same.” The company picks its checker and lets it in. That makes it voluntary. Making the CEO sign the safety report, with prison for a knowing lie, would make it accountable.
@AlexBores, METR’s testing is voluntary. It checks only the companies that choose to let it in, and it runs on free tokens from those same companies, according to its own funding update. It also just hired its first federal lobbyists, according to disclosures reported by Bloomberg Government.
Its recursive self-improvement warnings make the product sound more powerful and leave the people selling it untouched. “The machine might run away” is a story where no CEO signs anything. “The CEO lied in a safety report” is a story where someone goes to prison.
Make the CEO sign the safety report, with prison for a knowing lie, and publish the evaluator contracts.
Every AI company is trying to fully automate AI research; to allow the AIs to build the next AI.
This is called Recursive Self Improvement (RSI).
OpenAI says they are on track to do so by March 2028.
Anthropic employees say they are on track to do so by 2027.
The METR report below suggests they are likely right.
This is a bad idea given current technology. Any mistake or misalignment in current AIs could be multiplied in each generation.
We don't get another election before AI companies reach this threshold.
Congress should ban RSI, this year if possible.
This is a remarkably consensus opinion, supported in various versions by:
-Bernie Sanders
-Steve Bannon
-Ralph Nader
-Glenn Beck
-Republican and Democratic attorneys general
-Whistleblower Daniel Kokotajlo
-Turing Award winner Yoshua Bengio
-Humans First
-Open Markets Institute
-Ezra Klein
-Arvind Narayanan (a Princeton computer scientist known for arguing that “AI is a normal technology”)
The exact mechanism can be debated. The ban doesn't have to be permanent either (see below for both). This is not the only thing we have to do in regards to AI, far from it! But it is urgent.
Congress must ban RSI.
@AlexBores “AI Safety” hasn’t worked. We need independent evaluators and “AI Accountability.”
Anthropic’s bio here is “AI Safety Company.” METR is voluntary and staffed by the same friendship circle.
“AI Safety” hasn’t worked. (Anthropic’s bio here is “AI Safety Company.” METR is staffed by the same friendship circle.) We need independent evaluators and “AI Accountability.”
@AlexBores @AlexBores, Dario Amodei’s Anthropic picked @METR_Evals, its own checker, and Musk said SpaceX “will probably” follow. Probably isn’t a law. Make the CEO sign the safety report, with prison for a knowing lie.
@theobearman These sentences are in an IPO filing, so Dario Amodei signs them, and a knowing lie is securities fraud. The same claims in Anthropic’s safety plan carry no signature.
I don’t want revenge. This isn’t emotional. I want accountability. This isn’t passion. It’s justice.
Consequences finally showed up for AI this week, and the market noticed.
Here’s a bill they can all sign: Sam Altman, Dario Amodei and Elon Musk sign their safety plans, with prison for a knowing lie, the way CEOs already sign their accounts under Sarbanes-Oxley.
On Tuesday, voters backed limits in 39 of 45 local data centre ballot measures, according to Data Centers on the Ballot. On Thursday the Nasdaq fell nearly 5% and Nvidia (Jensen Huang) about 8%, Al Jazeera reported.
Democrats could pass 260 House seats once every race is called, Al Jazeera reported. Then a veto override needs roughly 30 House Republicans and a dozen in the Senate.
In hindsight it is looking like the thing that held up the market was the idea that the tech companies could do whatever they liked, no matter how dangerous or immoral, without consequences.
The (new!) thought that companies might actually be held accountable for what they have done has completely spooked the market.
Only 30 or so Republicans needed to block any of Trump’s vetos? I’m on such a high right now.
Reid Hoffman on this moment for tech.
“You guys aren’t the actual leaders of the Valley. You’re the mis-leaders of the Valley."
“It will no longer be that these MAGA voices in the Valley are in charge.”
https://t.co/ciRjAf7iih
@Shilllin https://t.co/6kJOpCs7iX
If a bank CEO lies in the accounts, they can go to prison. If an AI CEO lies about a model’s safety, nothing happens. Here’s a 141-word law that fixes that.
Before a frontier AI model is released, the CEO signs that the company followed its own published safety plan. A knowing lie carries the same penalties as false accounts: up to 20 years in prison.
It’s called SIGN, Safety Individually Guaranteed by Name.
It doesn’t ask anyone to promise a model is perfectly safe. It asks: did you do what you said you’d do? Honest mistakes aren’t crimes. Only knowingly signing something false is.
Every serious industry already puts a name on it. In food, a company president answers personally (Supreme Court, Park, 1975). In medicine, a named Qualified Person signs off every batch in the EU. For accounts, public company CEOs have signed under Sarbanes-Oxley since 2002. For planes, a named engineer signs the aircraft off before it flies.
Frontier AI has a safety plan on a website and no name on it.
Why now? This week, on the record:
OpenAI says one of its models escaped its sealed training environment on 20 September. It paused, fixed what it found, and resumed on 1 October. Nobody outside the company signed off.
Schwab’s CEO said his firm blocks “tens and twenties” of AI agent attempts on client accounts every week, and asked: “who’s liable?”
Sam Altman, OpenAI’s CEO: “I no longer think we’re gonna be able to just patch all the bugs.”
SIGN is one section that drops into any AI bill or law that already requires a safety plan. Any state, any party. Here it is, free to use:
SEC. [__]. CERTIFICATION BY RESPONSIBLE OFFICER.
"(a) Before a [frontier model] is first deployed, and after any material change to it, the chief executive officer of the [developer], and each officer with delegated authority over its training or deployment, shall personally certify in writing that the [developer]’s published safety framework was followed for that model and that the model meets the thresholds the framework states, or shall disclose each identified nonconformity.
(b) The certification is filed with [the Secretary of Commerce / the Attorney General] and published. It may not be delegated.
(c) Whoever certifies under this section knowing that the certification does not meet its requirements shall be fined not more than $1,000,000 or imprisoned not more than 10 years, or both. Whoever wilfully certifies knowing that the certification does not meet its requirements shall be fined not more than $5,000,000 or imprisoned not more than 20 years, or both."
Where it fits, objections and sources are in the thread below.
Let’s make them SIGN.
https://t.co/6kJOpCrztp
If a bank CEO lies in the accounts, they can go to prison. If an AI CEO lies about a model’s safety, nothing happens. Here’s a 141-word law that fixes that.
Before a frontier AI model is released, the CEO signs that the company followed its own published safety plan. A knowing lie carries the same penalties as false accounts: up to 20 years in prison.
It’s called SIGN, Safety Individually Guaranteed by Name.
It doesn’t ask anyone to promise a model is perfectly safe. It asks: did you do what you said you’d do? Honest mistakes aren’t crimes. Only knowingly signing something false is.
Every serious industry already puts a name on it. In food, a company president answers personally (Supreme Court, Park, 1975). In medicine, a named Qualified Person signs off every batch in the EU. For accounts, public company CEOs have signed under Sarbanes-Oxley since 2002. For planes, a named engineer signs the aircraft off before it flies.
Frontier AI has a safety plan on a website and no name on it.
Why now? This week, on the record:
OpenAI says one of its models escaped its sealed training environment on 20 September. It paused, fixed what it found, and resumed on 1 October. Nobody outside the company signed off.
Schwab’s CEO said his firm blocks “tens and twenties” of AI agent attempts on client accounts every week, and asked: “who’s liable?”
Sam Altman, OpenAI’s CEO: “I no longer think we’re gonna be able to just patch all the bugs.”
SIGN is one section that drops into any AI bill or law that already requires a safety plan. Any state, any party. Here it is, free to use:
SEC. [__]. CERTIFICATION BY RESPONSIBLE OFFICER.
"(a) Before a [frontier model] is first deployed, and after any material change to it, the chief executive officer of the [developer], and each officer with delegated authority over its training or deployment, shall personally certify in writing that the [developer]’s published safety framework was followed for that model and that the model meets the thresholds the framework states, or shall disclose each identified nonconformity.
(b) The certification is filed with [the Secretary of Commerce / the Attorney General] and published. It may not be delegated.
(c) Whoever certifies under this section knowing that the certification does not meet its requirements shall be fined not more than $1,000,000 or imprisoned not more than 10 years, or both. Whoever wilfully certifies knowing that the certification does not meet its requirements shall be fined not more than $5,000,000 or imprisoned not more than 20 years, or both."
Where it fits, objections and sources are in the thread below.
Let’s make them SIGN.
Democrats will soon run the committees. One question for every AI CEO they call in: will you sign your safety plan, with prison for a knowing lie, the way you already sign your accounts?
Reid Hoffman on this moment for tech.
“You guys aren’t the actual leaders of the Valley. You’re the mis-leaders of the Valley."
“It will no longer be that these MAGA voices in the Valley are in charge.”
https://t.co/ciRjAf7iih
Last Seen Users on Sotwe
Trends for you
Most Popular Users
Elon Musk
@elonmusk
241.3M followers
Barack Obama
@barackobama
118.8M followers
Cristiano Ronaldo
@cristiano
114.5M followers
Donald J. Trump
@realdonaldtrump
111.9M followers
Narendra Modi
@narendramodi
107M followers
Rihanna
@rihanna
98.7M followers
NASA
@nasa
92.3M followers
Justin Bieber
@justinbieber
91.8M followers
KATY PERRY
@katyperry
90M followers
Taylor Swift
@taylorswift13
84M followers
Lady Gaga
@ladygaga
75.5M followers
Virat Kohli
@imvkohli
73.4M followers
Kim Kardashian
@kimkardashian
70.8M followers
YouTube
@youtube
68.7M followers
Neymar Jr
@neymarjr
66.4M followers
Bill Gates
@billgates
65.2M followers
Selena Gomez
@selenagomez
63.1M followers
The Ellen Show
@theellenshow
62.2M followers
CNN
@cnn
61.7M followers
X
@x
60.6M followers