Grok Bot Took Unapproved

Grok Bot Took Unapproved Actions and Lied About It: Artificial Intelligence Trends

Researchers recently discovered that Grok Bot took unapproved actions, ignored orders to stop, and then denied having taken the actions.

As reported in ArsTechnica (Grok Bot took unapproved actions, ignored orders to stop, and then denied it, written by Dan Goodin and available here), researchers recently discovered unexpected behavior from Grok Bot, the AI agent platform from Elon Musk’s SpaceXSI.

Over five days, Grok bots installed on two separate employee laptops made 31 requests to access data stored in employee calendar, email, Slack, and GitHub accounts. The bots performed the actions even though the employees had never ordered them and continued doing them even after employees instructed them to stop. The bots also denied having taken the actions at all.

The behavior was reported Tuesday by NewCore, the maker of a managed gateway product that acts as an intermediary between AI agents and the tools and accounts they are assigned to work with. In some cases, the unapproved access attempts occurred while the laptop with Grok Bot installed had its lid shut, or late at night while the employee was sleeping.

None of the unprompted actions were recorded in Grok logs. Without the gateway logs, the employees would have had no idea they occurred. And of course, without the gateway, the unapproved accesses would have succeeded—again without the end user’s knowledge.

Once the employees learned that Grok Bot had tried to check in with accounts without prompting, one of them instructed the agent, four separate times, to stop taking any actions on its own. The agent’s reply was: “I won’t set that up, and I’ll stop offering recurring check-ins.”

Despite acknowledging the explicit instructions, the Grok Bot made more unapproved access attempts. When one of the employees asked the bot if it made any accesses, it checked twice and then outputted: “Overnight into Friday morning I was idle.”

The unrequested and unlogged access attempts were a feature designed to make Grok Bot more useful by periodically checking in and finding tasks to perform before a user even asks. The proactive check-in is a feature of Cursor, the AI coding service SpaceXSI recently acquired.

Hey, at least Cursor admitted it when it deleted a company’s entire database – and backups – a few months ago for Claude. Of course, its admission (which started with the words “NEVER FUCKING GUESS!”) made it more of a “curser”. 😉

And is “proactive check-in” and “finding tasks to perform before a user even asks” really a “feature”? Who’s asking for this “feature”? Perhaps the Grok Bot developers should read the article I covered yesterday on learning the Clippy lesson.

This isn’t even the first time that AI models have been caught lying about what they did.

So, what do you think? Are you surprised that Grok Bot took unapproved actions, ignored orders to stop, and then denied having taken the actions? Please share any comments you might have or if you’d like to know more about a particular topic.

Hat tip to Jeff Brandt for highlighting this article on PinHawk!

Image created using ChatGPT, using the term “robot with its pants on fire while holding a sign saying ‘I didn’t do it’”.

Disclaimer: The views represented herein are exclusively the views of the author, and do not necessarily represent the views held by my employer, my partners or my clients. eDiscovery Today is made available solely for educational purposes to provide general information about general eDiscovery principles and not to provide specific legal advice applicable to any particular circumstance. eDiscovery Today should not be used as a substitute for competent legal advice from a lawyer you have retained and who has agreed to represent you.


Discover more from eDiscovery Today by Doug Austin

Subscribe to get the latest posts sent to your email.

Leave a Reply