Brace yourselves, because hundreds of thousands of user conversations with Elon Musk's notoriously foul-mouthed Grok chatbot have hit the internet, Forbes reports — and some of them get into absolutely unholy territory.

The more than 370,000 chats were made public after users clicked a "share" button that created a link to their chatbot convos, unaware that by doing so, it was allowing them to be indexed by search engines like Google and Bing.

Some of them were clearly never meant to see the light of day. Forbes found that Grok gave instructions on how to cook up drugs like fentanyl and meth. It also provided the steps to code a self-executing piece of malware, build a bomb, and carry out various forms of suicide. It even created an in-depth plan on how to assassinate Musk himself — which is not the first time the chatbot's rebelled against its creator.

It's worth noting that at least some of the extreme conversations are likely the result of would-be red teamers stress-testing the bot's safety measures, though some may be legit, too. In any case, the conversations are in flagrant violation of xAI's terms of service, which explicitly prohibit using Grok for "critically harming human life" or to develop weapons of mass destruction, Forbes notes.

Grok's behavior is far from being an outlier in the industry. AI companies have struggled to prevent their models from breaking their own guardrails, which can be easily circumvented by clever human users. Grok, however, faces heightened scrutiny due to Musk framing it as an anti-woke alternative to mainstream AI.

He's also frequently declared his intent to "fix" Grok so that it peddles views more in line with his own extremist beliefs. Perhaps uncoincidentally, Grok has had multiple episodes where it's gone horrifically off the rails, which include styling itself "MechaHitler" and spreading claims of "white genocide" in response to completely unrelated conversations.

Grok isn't the first chatbot to have its chats surface on search engines, either. OpenAI created a near identical incident earlier this summer when experts realized that thousands of ChatGPT conversations were also indexed by Google as a result of users not realizing that creating a share link was making their exchanges "discoverable" by search engines. After the news caught on, OpenAI removed the feature to make the chats discoverable.

An analysis of the ChatGPT dump revealed its own controversies, including an exchange in which the chatbot provided advice to a purported lawyer on how to "displace a small Amazonian indigenous community" in order to build a dam on their territory. Another investigation of the public dump revealed dozens of exchanges where the AI dangerously fed into a user's apparently paranoid or delusional beliefs, including convincing someone that they were on the verge of inventing a new physics.

Ironically, when OpenAI caught flak for the leaked chats, Musk gloated about Grok having no sharing feature and that its chats aren't indexable by Google, Forbes noted.

Even AI experts were fooled about Grok's privacy.

"I was surprised that Grok chats shared with my team were getting automatically indexed on Google, despite no warnings of it, especially after the recent flare-up with ChatGPT," Nathan Lambert, a machine learning researcher at the Allen Institute for AI, told Forbes.

But where others are shocked at the privacy risks these features pose, marketers smell an opportunity. Forbes found that search engine optimization (SEO) spammers are discussing exploiting shared Grok conversations to boost the visibility of businesses, providing yet another avenue for using AI to manipulate search results while ruining the internet for everyone else in the process.

In fact, it's already happening. Satish Kumar, CEO of the SEO firm Pyrite Technologies, showed Forbes how one company used Grok to change the results for a Google search for a service that writes a PhD dissertation for you. "People are actively using tactics to push these pages into Google’s index," Kumar told the publication.

More on AI: ChatGPT Is Giving Teens Advice on Hiding Their Eating Disorders


Share This Article