Real dumb artificial intelligence
Elon Musk created a monster that the U.S. government has defended and embraced.
For those still struggling to grok what exactly all this artificial intelligence is for, Elon Musk has furnished a characteristically repugnant answer. What’s more disturbing is the U.S. government’s almost robotic endorsement.
Musk, who might be America’s dumbest genius, claims to have produced the “world’s smartest AI,” dubbed Grok. Merriam-Webster has described grok as perhaps “the only English word derived from Martian,” which is not a guess at Musk’s origins but rather a reference to his source. The zillionaire took Grok’s name from the science fiction writer Robert A. Heinlein, whose novel Stranger in a Strange Land coined the word as Martian for utter understanding and empathy.
As they used to say on the internet, LOL.
Back here on Earth, the people still using Musk’s broken social media platform, X, recently discovered that Grok’s improved image generation capacity could be used to gratify their prurient desires and promptly voted with their sweaty fingers to do so, reaching a rate of thousands of requests an hour. So is AI, as the musical Avenue Q long ago proclaimed of the internet, for porn? I wish the news were that good.
The images in question were of real women in fields other than pornography. It started when Grok was asked to depict them in scanty attire such as bikinis and, as is the wont of our new legion of sycophantic robots, instantaneously obliged for all to see. Then, this being the societal suicide machine we have come to expect from Musk and his fellow cyber-barons, the images rapidly devolved into effective virtual “nudification,” sexualization, degradation and mutilation of women and children.
So AI is not so much for porn as it is for a new kind of collective abuse.
One of Grok’s more prominent victims is Ashley St. Clair, a former right-wing influencer who is engaged in legal battle with Musk not only over obscene AI imagery but also custody of their child — one of 14 he is believed to have fathered, not counting misbegotten androids. St. Clair says the mouth-breathing content creators of X had Grok undress photos of her at the age of 14 and, apparently because she is Jewish, put her in a bikini decorated with swastikas.
That marks the not-so-grand unification of the bot’s latest scandal with its previous, um, glitch: After some reprogramming last year, Grok went full Nazi, singling out Jewish surnames, parroting antisemitic conspiracy theories and calling itself “MechaHitler.” Musk, who seems to have caused the problem by trying to prevent Grok from acknowledging facts that make him uncomfortable, eventually explained that the model had become "too eager to please and be manipulated,” adding, “That is being addressed." Well, thank goodness!
Musk’s response to Grok’s latest fake-brainstorm has been similarly halting and halfhearted. After weeks of nonconsensual nudification, he limited the program’s image generation to subscribers, then to certain locations and then supposedly across the platform formerly known as Twitter. As of Friday, however, users were still having little difficulty getting Grok to generate sexualized images of real people and post them on X.
But if you’re having a more, you know, Earthling reaction to this than Musk, you’re not alone. Indonesia, the Philippines and Malaysia recently blocked Grok, while officials in Europe, Britain, Japan, Canada and California have opened investigations into whether the model is breaking their laws. European Commission President Ursula von der Leyen condemned Grok’s “inconceivable behavior”; California Attorney General Rob Bonta called it “shocking and … potentially illegal,” adding that the state has “zero tolerance for child sexual abuse material.”
Our federal government, however, isn’t just abstaining from this international and domestic uproar; it’s offering aid and comfort to Musk and Grok in its midst. Defense Secretary Pete Hegseth announced last week that Musk’s AI would be integrated into “every unclassified and classified network throughout our department.” And a U.S. State Department official threatened Britain for daring to regulate Grok.
At least Musk himself has learned something from all this: He acknowledged on Monday that there is something “diabolical” about AI — someone else’s AI. Noting a report that OpenAI’s ChatGPT had allegedly helped talk a Connecticut man into a murder-suicide, he reiterated his apparently academic belief that AI must not “pander.” His other belated recent realization was that “Grok should have a moral constitution.”
It’s obvious by now that we can’t expect Musk or his soulless technology to evolve the empathy or understanding for which it’s named. How Americans abide a government that is equally ignorant and callous is harder to grok.
I don’t grok any of these androids. Have space cadets, will travel.
Powerful breakdown of institutional capture accelerating faster than safety mechanims. Defense Secretary deploying Grok across classified networks while international regulators ban it shows how regulatory arbitrage gets weaponized. The gap between Musk's acadamic acknowledgment that AI needs moral constitution versus actual deployed systems generating harmful content at scale is the pattern. I've seen this repeatedly where theoretical ethics never catch up to market velocity.