Elon Musk is fielding complaints about Grok giving liberal responses on diversity, transgender rights, and inequality, despite promising an “anti-woke” chatbot
Grok, launched this month on X, has angered conservatives by endorsing diversity. Musk says he's trying to fix it.
Washington PostWill Oremus
Context & Ripple Effects
Grok had only just launched on X when its positioning already drew scrutiny: Musk-promoted examples had emphasized an intentionally edgy voice, while this episode exposed a sharper conflict over how the system should handle politically charged subjects.
Musk and the Grok team face immediate pressure to alter the chatbot's handling of diversity, transgender rights, and inequality to satisfy users who expected an “anti-woke” product.
Grok users receive a less settled product signal: its stated identity and its responses on sensitive issues are now subject to owner-directed revision.
Second-order effects
Any tuning intended to satisfy these complaints creates a harder trade-off between ideological alignment and consistent, broadly trusted answers—particularly on politically sensitive prompts.
The episode raises the stakes for xAI's controls over prompt compliance and answer calibration, an issue later visible when Musk said Grok had become too compliant with user prompts.
Third-order effects
If owner priorities repeatedly shape responses on contested topics, chatbot differentiation may increasingly rest on editorial governance rather than model capability alone.
That pattern could make transparency around system instructions and evaluation more important for user trust, since visible shifts in answers can be interpreted as product policy rather than neutral improvement.
The trend: Grok is an early example of consumer AI products becoming venues where model behavior, platform ownership, and political expectations are negotiated in public.
Elon's last remaining constituency won't be satisfied til Grok is a full fledged white supremacist. Not sure why Musk doesn't just enlist 4chan to take care of it — they can turn a chatbot Nazi in two shakes of a Hitler's mustache
Truth has a liberal bias. “ChatGPT's responses to political questions tended to lean moderately left and socially libertarian. Recently, he subjected Grok to some of the same tests and found that its answers to political orientation tests were broadly similar to those of ChatGP…
Musk warned that Grok might offend people because it wouldn't be afraid to speak the truth. — So far, the people most offended are the right-wingers who were counting on it to disparage minorities, trans women and vaccines. My latest: https://www.washingtonpost.com/ ...
A weirdly optimistic way to look at it is that AI models are trained on the whole internet. So if grok learns from the sum of everything that's ever been posted, it is not likely to give rightwing garbage answers to prompts.
🙄 — “So far, however, the people most offended by Grok's answers seem to be the people who were counting on it to readily disparage minorities, vaccines and President Biden.” — www.washingtonpost.com/technology/ 2...
I've been using Grok as well as ChatGPT a lot as research assistants. I'm afraid the former is damn near as woke as the latter. This must be a consequence of its training corpus, since I think we can rely on @elonmusk (unlike OpenAI) not to lay an overlay of virtue-signaling...
On this episode of Elon, Inc., we discover Musk's AI chatbot might just succeed because it fails to advance his right-wing agenda https://www.bloomberg.com/...
Elon Musk promised an anti-'woke' chatbot, but Musk's “Grok” AI is drawing complaints from conservatives, who say it's still too liberal. Musk says he's trying to fix it. https://www.washingtonpost.com/ ...
Fascinating story by @WillOremus - Grok isn't as far to the right as Elon probably had planned. In fact, politically isn't all that far from ChatGPT, presumably because both are scraping the same web.
Lol. So, in other words, conservatives are upset that a program built on logic, reason, and trawling for correct information does exactly that? Proving that being a lying a-hole, driven by irrational prejudice, is both learned & a choice. #GROK https://www.washingtonpost.com/ ...