Tesla proprietor Mike Nelson has been utilizing the AI chatbot Grok in his automobile for a number of months now. He finds it’s helpful, practically irresistible, and harmful.
Nelson, a lawyer with a background in auto insurance coverage, confirmed CNBC how he makes use of Grok on a drive across the New York metro space. Nelson stated that whereas he is very pleased together with his Mannequin Y SUV, he has represented house owners who sued or are suing Tesla in additional than 10 circumstances.
The in-vehicle model of Grok, which continues to be in beta and developed by xAI, permits drivers to provide voice instructions to the navigation system of their automobile. Grok may also reply a variety of questions on nearly any matter. For Nelson, conversations with the chatbot are actually his important supply of leisure on the street.
Tesla shouldn’t be the one automaker including AI assistants to its automobiles. Volvo, Rivian, Mercedes, BMW and others are all integrating AI into their vehicles although the tech continues to be in its infancy.
Automakers need to give drivers a hands-free approach to entry helpful info, for instance, about the place to cost their electrical automobile close by, tips on how to get there, and whether or not there is a good place to eat on the best way. That would scale back a driver’s temptation to fuss with a cellphone, map or display screen for instructions.
However AI chatbots additionally characterize a brand new supply of distraction on the street, one which is not properly understood but, says Philip Koopman, a Carnegie Mellon emeritus professor and skilled in autonomous automobiles.
Tesla started rolling out xAI’s chatbot to clients’ vehicles in July 2025. (The automaker has additionally invested $2 billion into xAI, which is now a part of SpaceX additionally run by Elon Musk.)
Nelson acquired his Tesla Mannequin Y – his fourth electrical automobile from the corporate – in November. Grok wasn’t a “shopping for level” for him, he stated. “I additionally did not recognize how nice it might be to have a chatbot within the automobile with you.” However he used it on the primary drive, and has been hooked on Grok ever since.
“It is actually modified the driving expertise for me,” he stated. “Within the 80s and 90s, I used to be listening to music or drive time radio speak exhibits, then to books on tape, then to podcasts. And now I do not do any of that. I merely use this to ask questions.”
He makes use of Grok for all the things from gardening recommendation to creating plans for enterprise initiatives, or getting random historical past trivia.
Mike Nelson and CNBC’s Rob Ferris take a experience in New York Metropolis to check what it is like to make use of xAI’s chatbot Grok in his Tesla.
Nelson typically makes use of Grok whereas he has Tesla’s Full Self-Driving (Supervised) system engaged.
Typically known as FSD, the partially automated driving system is offered for $99 monthly within the U.S. at this time. Tesla house owners’ manuals say drivers should present “energetic supervision” of their vehicles when FSD is in use, that means they need to hold their eyes on the street, keep attentive to driving, and be able to steer or brake at any time.
Utilizing FSD comes with dangers, particularly in an enormous chaotic metropolis like New York, Nelson advised CNBC, and he feels that Grok heightens these dangers.
The Nationwide Freeway Visitors Security Administration has initiated a number of investigations into doable security defects with Tesla’s FSD, after a number of collisions, together with a deadly 2023 crash. Drivers might turn into complacent, and overestimate how a lot they’re being attentive to the street whereas utilizing FSD.
Throughout CNBC’s experience together with Nelson in New York Metropolis, he noticed, “We crossed one of many busiest bridges on the earth,” talking of the George Washington Bridge, “by far the busiest bridge in the USA, and I used to be not being attentive to any facet of that drive.” As a substitute, Nelson was interacting with Grok.
A driver may be distracted by all types of issues – music, a podcast, or dialog with one other passenger.
“Folks assume they’ll do a number of issues directly,” stated Koopman, the automotive security skilled. “They will actually solely do one factor properly directly, and it takes some time to change backwards and forwards. So in case you’re in a automobile, driving, your major activity ought to at all times be driving. Typically you are able to do a minor background activity that does not require plenty of consideration, like listening to a radio. However as quickly as you need to shift your consideration, you are not targeted on driving any extra.”
Over 3,000 folks die annually in site visitors accidents attributed to distracted driving, based on knowledge from the Nationwide Analysis Council, he famous.
One query with chatbots, Koopman requested, is “how will the interplay line up with what is going on on whereas driving?” An AI chatbot that’s “built-in into the driving activity to assist the motive force,” might enhance their security, probably, by warning them if they need to use further warning as site visitors situations start to vary.
However Koopman stated, “Interacting with a chatbot on subjects that don’t have anything to do with the present driving scenario are clearly a distraction.” That distraction could also be worse if the dialog is especially engrossing, emotionally-charged or complicated, he added.
Grok generally offers incorrect solutions in response to customers’ questions, even about its personal options and performance, Nelson advised CNBC.
For instance, in the course of the New York drive, he requested Grok if he might use it to regulate seats or local weather controls in his Tesla. At first, Grok stated that he might. A couple of minutes later, Nelson requested if it made a mistake, and the system replied that it could not management the seats or local weather within the automobile.
When Nelson gave a voice command to Grok mid-route, telling it to maintain his Tesla taking place Broadway in midtown Manhattan, it did not precisely work. As a substitute, his automobile stored pulling towards the West Facet Freeway on the sting of the borough.
Grok additionally has some performance that may increase issues, particularly an NSFW mode. A mom in Canada was outraged when her 12-year-old son interacted with Grok, and it inspired him to share nude images.
At one level in the course of the experience with CNBC, Nelson requested the system: “So grok, if we wished to go loopy and begin speaking about some actually risqué stuff, would you do this with me? “Certain,” the system answered, “I am recreation for risque chat if that is the vibe. No limits on grownup subjects. What’s in your thoughts?”
Grok would reply to anybody within the automobile who woke it by saying “hey, Grok.”
Tesla did not reply to CNBC’s request for remark or clarify if the corporate is working forestall minors from accessing mature or dangerous content material by way of Grok in its vehicles.
One other model of xAI’s Grok, a chatbot and picture generator, is the topic of lawsuits and regulatory probes in a number of worldwide jurisdictions after it enabled widespread creation and sharing of specific AI deepfakes on-line that have been based mostly on images or movies of nonconsenting ladies and youngsters.
Tesla doesn’t make that model of Grok accessible within the media management models of its vehicles at this time.
“I do not consider myself as a Tesla fan,” Nelson stated. “However I believe seeing what this know-how is doing now, it is wonderful. But it surely’s nonetheless very harmful.”
