Exploring latest planets implies that you’re all the time operating prior to now. It takes about twenty minutes for a signal to achieve a Mars rover from Earth; by the point a brand new instruction arrives, the rover will have already got acted on the previous one.
But on December 8 and 10, 2025, the commands that were sent to NASA’s Perseverance Rover looked like something from the long run. That’s because, for the primary time ever, they’d been written by an AI.
Specifically, they were written by Anthropic’s AI model, Claude. Engineers at NASA’s Jet Propulsion Laboratory (JPL) used Claude to plot out the route for Perseverance to navigate an roughly four-hundred-meter path through a field of rocks on the Martian surface.
Due to delay within the signal to the rover, operators can’t micromanage where it drives. They plan a route, send it, and only later see the outcomes. Until now, human experts have all the time been those to do this planning. This time, Claude lent a hand.
4 hundred meters isn’t far: it’s one lap of a running track. Nevertheless it’s a start. Claude—the identical AI model that folks use to draft emails, construct software apps, and analyze their company’s funds—is now helping humanity explore other worlds.
The Perseverance Rover—a car-sized robot bristling with cameras and scientific equipment—has been energetic on Mars because it landed in February 2021. Its mission is to characterize the planet’s geology and past climate, collecting samples of Martian rock and regolith (broken rocks and mud) and paving the best way for human exploration of the Red Planet. One other of its key objectives is astrobiological in nature: its landing site, the forty-five-kilometer-wide Jezero crater, was chosen due to the evidence that after, billions of years ago, it contained water—water that may need supported microbial life. To date, the rover has discovered tantalizing hints of what could be ancient biology on Mars.
But driving on the Martian surface is hardly trivial. Every rover drive must be fastidiously planned, lest the machine slide, tip, spin its wheels, or get beached. So ever for the reason that rover landed, its human operators have painstakingly laid out waypoints—they call it a “breadcrumb trail”—for it to follow, using a mix of images taken from space and the rover’s onboard cameras. Once it’s drawn up, the plan is transmitted across the 362 million kilometers between Earth and Mars using the Deep Space Network. Having received the signals, the rover embarks on its trip.
That is high-stakes work. In 2009, the Spirit rover, considered one of the forebears of Perseverance, drove right into a sand trap and never moved again.
All this meticulous planning is time-consuming. Perseverance has an AutoNav system that helps it steer around obstacles between the waypoints, however it only sees things from the rover’s own perspective and might’t plan too far ahead.
JPL’s engineers tested whether Claude could save them a few of that laborious work by helping to plan Perseverance’s route—and accomplish that with the identical level of accuracy as a human operator. They arrange a process in Claude Code to delegate the waypoint-setting to the AI.
Claude didn’t do that with a single prompt. As an alternative, the model needed context before it could effectively plot the waypoints. The JPL engineers gathered together the info and experience they’d gained from years of driving the rover, and provided it to Claude Code. With all this extra information, Claude used its coding skills to put in writing commands in Rover Markup Language—the bespoke, XML-based programming language originally developed for the Mars Exploration Rover mission.
Using its vision capabilities to research the overhead images, Claude planned Perseverance’s breadcrumb trail point by point for sol 1707 and sol 1709 (a sol is a Martian day; these were the near-equivalent of December 8 and 10 on Earth). It strung together ten-meter segments right into a path, then iterated to refine the waypoints—critiquing its own work and suggesting revisions.
As with all AI output, it’s essential to envision Claude’s work. The waypoints drawn by Claude were run through a simulation that Perseverance uses daily to verify the accuracy of the commands: over 500,000 variables were modeled to envision the projected positions of the rover and predict any hazards along its route.
When the JPL engineers reviewed Claude’s plans, they found that only minor changes were needed. For example, ground-level camera images (which Claude hadn’t seen) gave a clearer view of sand ripples on either side of a narrow corridor; the rover drivers elected to separate the route more precisely than Claude had at this point. But otherwise, the route held up well. The plans were sent to Mars, and the rover successfully traversed the planned path.
Over 500,000 variables were modeled to envision the projected positions of the rover and predict any hazards along its route.
The engineers estimate that using Claude in this manner will cut the route-planning time in half, and make the journeys more consistent. Less time spent doing tedious manual planning—and fewer time spent training—allows the rover’s operators to slot in much more drives, collect much more scientific data, and do much more evaluation. It means, in brief, that we’ll learn rather more about Mars.
Claude’s role within the Perseverance mission is in some ways only a test run for what comes next.
The sorts of autonomous capabilities that Claude showed on the Mars rover drive—quickly understanding novel situations, writing code to operate complex instruments, making smart decisions without an excessive amount of hand-holding from its operators—are exactly people who’ll prove useful on longer and more ambitious space missions.
NASA’s upcoming Artemis campaign goals to send humans back to the Moon, and to eventually establish a US-led base on the lunar south pole. Doing so will involve overcoming countless engineering challenges—and identical to on Mars, using resources efficiently might be the watchword.
Just as Claude can apply its intelligence to the range of somewhat more sublunary tasks we feature out on Earth, developing a general AI assistant that’s versatile and reliable enough to assist with all the things from mapping lunar geology to monitoring the astronauts’ life-support systems might be a force multiplier for NASA missions to the Moon and Mars.
NASA’s upcoming Artemis campaign goals to send humans back to the Moon, and to eventually establish a US-led base on the lunar south pole.
Even further in the long run, autonomous AI systems could help probes explore ever more distant parts of the solar system. Such missions would present fiendish technical problems: solar energy would turn out to be less and fewer viable; the delay on sending signals from Earth could stretch to hours; and the pressure, temperature, and radiation of the destinations would conspire to render a robotic explorer’s lifetime far riskier—and much shorter.
Claude’s four-hundred meter drive on Mars provides the primary glimmer that we’d have the ability to unravel those problems, and construct a future filled with truly autonomous machines that could make fast, adaptive, efficient decisions without waiting for human input. A future where sooner or later our probes might visit moons like Europa or Titan, descend through their icy shells, and chart their very own course through the dark oceans below.
