Since 2021, NASA’s Perseverance rover has achieved several historic milestones, including the first return. audio recording from Mars. Now, almost five years have passed landed on the Red Planetit just achieved another success. Last December, Perseverance successfully completed a route through a section of the Jezero crater planned by Anthropic’s Claude chatbotwhich marks the first time NASA has used a large-scale language model to pilot a vehicle-sized robot.

Between December 8 and 10, Perseverance drove approximately 400 meters (about 437 yards) through a field of Martian surface rocks mapped by Claude. As you can imagine, using an AI model to plot a course for Endurance is not as easy as inputting an impulse.

As NASA explains, routing the Endurance is not an easy ​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​easyèบที่พื่พลียสลียลียลียยยบับ task, even for a single person. “Each rover drive must be carefully planned, perhaps the machine will slide, tip, spin its wheels, or beach,” says NASA. “So since the rover landed, its human operators have been painstakingly laying out waypoints – they call it the ‘breadcrumb trail’ – for it to follow, using a combination of images taken from space and the rover’s onboard cameras.”

In order for Claude to complete the task, NASA must first provide Claude CodeAnthropic’s programming agent, with “years” of contextual data from the rover before the model begins to write a route for Endurance. Claude then went through the mapping process methodically, combining waypoints from ten-meter segments that were later critiqued and iterated.

This is NASA we’re talking about, engineers from the agency’s Jet Propulsion Laboratory (JPL) make sure to double-check the model’s work before sending it to Endurance. The JPL team ran Claude’s waypoints through a simulation they use every day to confirm the accuracy of the commands sent to the rover. In the end, NASA said it only needed to make “minor changes” to Claude’s route, with one tweak coming as a result of the fact that the team had access to ground-level images that Claude didn’t see during the planning process.

“Engineers estimate that using Claude in this way will cut route planning time in half, and make flights more consistent,” NASA said. “Less time spent doing tedious manual planning — and less time spent training — allows rover operators to fit in more drives, collect more science data, and do more analysis. This means, in short, that we can learn more about Mars.”

Meanwhile the productivity gains offered by AI often overstatedin the case of NASA, any tool that allows its scientists to be more efficient is certainly welcome. In the summer, the agency lost about 4,000 employees — accounting for about 20 percent of its workforce — due to cuts by the Trump administration. Going into 2026, the president proposes to cut the agency’s science budget by almost half before Congress at last rejected that plan in early January. However, even preserving its funding below 2025 levels, the agency has a tough road ahead. It was asked to return to the Moon with the less than half of the workforce this was during the height of the Apollo program.

For Anthropic, on the other hand, it’s a big feat. You remember last spring Claude can’t even beat Red Pokémon. In less than a year, the company’s models have gone from struggling to navigate a simple 8-bit Game Boy game to successfully plotting a course for a rover on a distant planet. NASA is excited about the possibility of future collaborations, saying that “autonomous AI systems will help probes explore the more distant parts of the solar system.”