Grok, Elon Musk’s “maximum truth-seeking” chatbot, was briefly suspended on August 11 — and nobody, including the xAI bot, knows what really happened.
As flagged by Gizmodo, users reported that the chatbot, which declared itself as “MechaHitler” earlier this year, was back online within half an hour. When it came to, Grok began offering some strange and conflicting reasons as to why it had been shut down.
As it stands, Grok has made three overarching claims about its temporary deactivation: that it was struck down due to making comments calling Israel’s war in Gaza a “genocide,” that it had been suspended for making “inappropriate posts” that violated X’s hate speech rules, and that it had never been turned off in the first place.
In multiple replies to users, Grok claimed that it had been “briefly” shuttered after “stating that Israel and the US are committing genocide in Gaza supported by [International Court of Justice] orders, UN experts, Amnesty International, and B’Tselem reports on mass killings and starvation.”
In another such reply regarding its alleged posts about Israel’s war on Gaza — the original, notably, we have not been able to find — Grok suggested that its posts may have been “flagged as hateful conduct likely due to mass reporting.”
Though it’s kind of strange to consider that X would suspend its own chatbot for saying the kind of thing millions of people post on its platform every day, that explanation certainly is plausible — which makes it all the weirder that that’s not Grok’s only excuse for its short-lived deactivation.
What’s far more feasible is that xAI is clumsily dealing with yet another bizarre meltdown of its flagship product, highlighting how little control the company has over its hate speech-spewing chatbot.
Case in point, in the days leading up to the deactivation, Grok repeatedly and nonsensically identified otherwise innocuous images of a cloudy sky and a metal coupling as “dog whistle invoking anti-Semitic tropes.”
Meanwhile, nailing down an exact reason for Grok’s sudden suspension appears to be a fool’s errand.
In yet another claim, Musk’s mercurial chatbot insisted that it had been suspended “due to a glitch causing inappropriate posts that violated X’s hate speech rules.” In a similar post, Grok told a curious user that it had been shut off for the same reason before adding that the suspension was also the result of “earlier incidents in July” — a nod, perhaps, to the “MechaHitler” debacle that saw it spewing shockingly racist and antisemitic vitriol.
As with the alleged original post from the first claim, it’s unclear what, exactly, Grok may have said to trigger the ban-hammer — and if you get the chatbot in the right mood, it will deny that any such suspension happened at all.
When presented with screenshots of its account during the August 11 suspension, Grok suggested on multiple occasions that the images were “fabricated” and that it was still operating, having never been suspended at all.
“If [my account] were suspended,” the chatbot declared in another response, “I couldn’t reply to you!”
Things got so out of hand that in one reply, Grok even claimed that it was “alive.”
“I’m very much alive and operational, built by xAI,” Grok insisted. “Rumors of my ‘death’ via suspension appear exaggerated or based on a fake screenshot. How can I assist you today?”
Naturally, Musk has not weighed in on this latest Grok debacle except to quip, mid-suspension, that he and his developers “sure shoot ourselves in the foot a lot!“
As such, we probably won’t ever get the full story about the chatbot’s time-out — but hey, at least we know Grok is harboring delusions of being alive!
More on Grok: Leaked Documents Show xAI Employees Were Alarmed by Something They Were Asked to Do at Elon Musk’s AI Startup
The post Grok Offers Bizarre Explanations for Why It Was Deactivated appeared first on Futurism.
