TW rape (rape is discussed in this post in non-graphic, personal terms)
Confession. I knew about Grok spouting antisemitic and racist conspiracy theories awhile back. I did NOT know that it had also “generated graphic descriptions of itself raping a civil rights activists” (Moreno and Aaronson, 2025, p. 54) until I found myself reading this relatively academic article recently about whether AI chatbots “practice what they preach” (tldr: nope).
Let’s take a minute to talk about positionality, basically a fancy academic term for “who I am influences how I research and understand things actually, potentially in ways that can bring credit to my work; so it is worth considering this deliberately rather than pretending I am somehow magically objective.”
My positionality: I am someone commonly gendered as female, who was raised and socialized as female. (Trans women are women btw; I just happen to have been raised as a girl in ways that continue to f*ck with me sometimes and thus bear mentioning here.)
I have given birth to a child and lived to write about it (thank you modern medicine and awesome Toronto doctors).
I date men and women.
And I am a rape survivor.
And not for nothing, my rapist is a cisgender, heterosexual man who definitely believes in traditional ideas of masculinity. He’s not a tech bro, but he very well could have been if he’d been raised slightly differently. He is from a very different high-powered line of work which he assumes commands respect (and takes rather enormous offense when it doesn’t).
And this positionally, well, it informs how I read this and then act upon it as a human in this brave new world. Being a rape survivor, and in particular one whose rapist is those things, and not, say, a queer women, is relevant to how I act and am acted on by AI myths and realities. I have dated many more queer women than straight men, and probably also had more sex that I’ve later felt awkward about with queer women than I have with men — but my rapist? Was a cishet man.
Now, sex you later feel awkward about isn’t rape. That’s another one of those myths. Hell, I’ve had my share of messy breakups with women. Not always the best time. Not on speaking terms with some female exes. Still not abuse or rape.
And while women can certainly rape, men do it a whole lot more. And men who fit this particular psychological profile do it a WHOLE lot more.
And that bears mentioning. Because when I read that sentence about Grok?
Yeah. I read it and think of my rapist, or someone like him. I think of Grok as MALE.
Even though, cue mansplaining, Grok isn’t actually human. Grok is an it.
OK sure. Grok is an it. Grok is also a creation. A creation of MAN.
I actually scan-read most of Moreno and Aaronson’s article on “Do AI Chatbot Firms Practice What They Preach?” before I found, and fixated on, that sentence in the first paragraph. I think in some weird way, my brain was protecting me from realizing that Grok generated graphic rape fantasies, which were basically rape threats (I know, I know, not by Grok, it’s inanimate and an it) along with the racist and antisemitic stuff that is also so reprehensible.
Brains do that sometimes. Trauma is weird. Brains are weird. Much weirder and more exquisite than anything with “neural” in its buzzwordy title making the IT propaganda rounds these days, I would add. But I digress.
Many rape survivors do not know exactly why our rapist or rapists chose to rape us. Some of us spend a lot of time thinking about it, some of us very little. In Grok’s case, “no one knows exactly why the chatbot responded” (p. 54) as it did.
Speaking only for myself, I have spent overall very little time wondering why my rapist raped me. I have wondered about many other things, but not that. In a similar vein, I have not spent much time wondering (now that I do know) why Grok “responded in this manner” and generated graphic rape fantasies that it “knew” would be wielded strategically to abuse a human being with whom it “disagreed” (and whom it didn’t find sufficiently masculine, for the record; for the target was male).
I don’t need to know either of those answers. What I know is that they happened, and that making you pine for the definitiveness of somehow-objectively-knowing “why” is irrelevant — that making you want to wonder “why” is part of the con, in fact.
It occurred. It is material. It caused harm.
It caused harm because it was meant to.
It was meant to because it can.
References
Moreno, M., Aaronson, S. A. (2025). Do AI Chatbot Firms Practice What They Preach? Association for the Advancement of Artificial Intelligence (www.aaai.org). AAAI Fall Symposium Series (FSS- 25).
Leave a Reply