The paper asks whether technosignature searches should look beyond radio emissions for long-lasting changes that technological societies may make to planetary surfaces. Its central idea is that intentional modifications—such as construction, mining, habitation, and waste disposal—can create non-random patterns visible from orbit. Because human societies have left traces on Earth that persist for centuries or millennia, the authors argue that similar features could be sought on other rocky bodies in the Solar System. Their aim is to establish quantitative limits on the presence of such artifacts, not to claim that any have already been found. The proposed approach combines planetary imaging and other remote-sensing data with machine-learning methods. Optical imagery could be searched for unusual shapes or patterns. LiDAR and ground-penetrating radar could reveal buried or subsurface structures, while reflectance spectroscopy could identify materials with unusual chemical or mineralogical properties. The authors recommend developing two types of machine-learning searches in parallel. Unsupervised or self-supervised anomaly detection can identify features that differ from their surroundings without relying on examples of extraterrestrial artifacts. Supervised classification, by contrast, can use known human-made structures as training data. The paper cites terrestrial archaeological research showing that supervised methods can identify structures in satellite imagery. However, the authors caution that training on human examples could cause searches to miss unfamiliar technologies or encourage false interpretations. The opportunity is supported by the scale and resolution of existing planetary datasets. The Lunar Reconnaissance Orbiter’s Narrow Angle Camera has collected more than two million lunar images, amounting to petabytes of high-resolution imagery. The Mars Reconnaissance Orbiter’s HiRISE camera can achieve a best-case resolution of 30 centimeters per pixel, and its public catalog contains more than 92,000 images. The authors also describe an initial self-supervised anomaly-detection method developed for lunar imagery. It identified the Apollo 17 lander, a known human artifact, as well as scientifically interesting natural features, including recent impacts and volcanic pits. This result demonstrates the method’s ability to flag unusual objects, but it is not evidence of extraterrestrial technology. The rationale for these searches partly rests on the potential longevity of surface traces. On bodies with little tectonic or biological activity, artifacts could remain visible far longer than comparable traces on Earth; the paper cites estimates of up to hundreds of millions of years. Most Solar System surfaces, the authors note, have not been explored at a resolution sufficient to rule out artifacts. Potential search locations include asteroids, rocky moons, Mars, and Kuiper Belt objects. The paper also considers the speculative possibility that an earlier technological species existed in the Solar System, while acknowledging that geological processes would likely have erased traces on Earth or Venus. The authors conclude that searches for surface technosignatures are a promising, relatively low-cost addition to technosignature research. Existing archives make it possible to begin searching without waiting for new missions, and the work could also benefit planetary science by revealing unusual natural formations. However, the case remains theoretical: no candidate extraterrestrial surface technosignature or confirmed artifact has been reported, and no quantitative upper limits have been published. The authors do not estimate the probability that such artifacts exist, and the methods are vulnerable to false positives, false negatives, and assumptions about what artificial structures might look like. They therefore recommend combining supervised and unsupervised searches and, where possible, following up with independent data and in situ investigation.
- The paper proposes searching for long-lasting, non-random surface changes—such as construction, mining, habitation, and waste disposal—as possible technosignatures beyond radio emissions.
- Its goal is to establish quantitative limits on possible artifacts, not to claim that extraterrestrial artifacts have been found.
- Searches could combine optical imagery, LiDAR, ground-penetrating radar, and reflectance spectroscopy to detect unusual shapes, buried structures, or materials.
- The authors recommend pairing unsupervised or self-supervised anomaly detection with supervised classification trained on known human-made structures.
- Human-based training data may help identify familiar structures but could miss unfamiliar technologies or encourage false interpretations.
- Large existing datasets make searches feasible now: LRO’s Narrow Angle Camera has collected over two million lunar images, and Mars Reconnaissance Orbiter’s HiRISE catalog has over 92,000 images at best-case resolution of 30 centimeters per pixel.
- An initial self-supervised method flagged the Apollo 17 lander and unusual natural features, demonstrating anomaly detection—not evidence of extraterrestrial technology.
- On geologically quiet bodies, surface traces might persist for up to hundreds of millions of years; most Solar System surfaces have not been surveyed at sufficient resolution to rule out artifacts.
- Potential search targets include asteroids, rocky moons, Mars, and Kuiper Belt objects; geological activity may have erased traces on Earth or Venus.
- Surface technosignature searches could be a relatively low-cost addition to research and may also reveal scientifically valuable natural formations.
- No candidate extraterrestrial surface technosignature or confirmed artifact has been reported, and no quantitative upper limits have been published.
- The methods remain vulnerable to false positives, false negatives, and assumptions about artificial structures; the authors recommend independent data and, where possible, in situ follow-up.
Summary1. Scientists want to look for lasting changes on other worlds, like buildings, mines, or places where things were left behind.
2. They have not found any alien-made objects; they want to see what might be there and set limits on what could have been missed.
3. Pictures and other tools can help find odd shapes, things underground, or unusual materials.
4. Computers can learn to spot strange things, but they might mistake natural features for objects or miss things that look unfamiliar.
5. Many places have not been checked closely, so scientists want to search them carefully and double-check anything unusual.
Definitions- Lasting: staying for a long time.
- Tool: something that helps people do a job.
- Unusual: not common or ordinary.
- Natural: made by nature, not by people.
- Double-check: look again to make sure something is right.
Looking Beyond Radio for Signs of Technology in the Solar System
Searches for extraterrestrial intelligence often focus on radio signals. But a technological society might leave evidence in other ways—by changing the surfaces of planets, moons, or asteroids. Mining, construction, habitation, and waste disposal could create patterns that differ from natural landscapes and persist long after their builders are gone.
A recent research paper explores how scientists might search for these surface technosignatures using planetary images and other remote-sensing data. The authors do not claim to have found alien artifacts. Instead, they propose ways to look for them and, eventually, set quantitative limits on how common such traces might be.
Why Look for Surface Traces?
On Earth, human activity has left structures and landscape changes that can endure for centuries or millennia. On worlds with little tectonic activity or biological growth, traces could last much longer. The paper cites estimates that some artifacts might remain detectable for as long as hundreds of millions of years.
That possibility makes the search relevant not only to current extraterrestrial life, but also to the more speculative idea that a technological species may once have existed somewhere in the Solar System. Geological forces would likely have erased many traces on Earth or Venus, but other rocky bodies may preserve their surfaces for longer.
Most Solar System terrain has not been surveyed at the resolution needed to rule out artifacts. The authors identify asteroids, rocky moons, Mars, and even Kuiper Belt objects as potential places to investigate. For now, the opportunity is less about knowing where technology might be and more about systematically checking the data already available.
Searching with Existing Planetary Data
Planetary missions have produced enormous image collections. The Lunar Reconnaissance Orbiter’s Narrow Angle Camera has collected more than two million lunar images, representing petabytes of high-resolution imagery. On Mars, the HiRISE camera aboard the Mars Reconnaissance Orbiter can reach a best-case resolution of 30 centimeters per pixel, and its public catalog includes more than 92,000 images.
These archives could be examined for shapes, layouts, or surface patterns that are difficult to explain through known natural processes. Other remote-sensing tools could add different kinds of evidence:
Optical images could reveal unusual geometric shapes, repeated patterns, or altered terrain.
LiDAR and ground-penetrating radar could help detect buried or subsurface structures.
Reflectance spectroscopy could identify materials with unusual chemical or mineralogical properties.
No single observation would establish that a feature is artificial. An unusual shape might turn out to be a crater, a volcanic formation, or an image-processing artifact. The goal is to identify candidates for further study, not to treat every anomaly as evidence of technology.
Two Complementary Machine-Learning Strategies
The authors recommend developing two approaches in parallel. The first is unsupervised or self-supervised anomaly detection. Rather than learning what an alien artifact looks like, a model searches for features that stand out from their surroundings. This is useful because extraterrestrial technology, if it exists, may not resemble familiar human structures.
The second approach is supervised classification, which trains models on labeled examples of known human-made structures. Archaeological research on Earth has shown that supervised machine learning can identify features in satellite imagery. Such methods could help locate familiar signatures of construction or land use in planetary datasets.
Each strategy has limits. A model trained on human examples could overlook unfamiliar forms of technology. Meanwhile, an anomaly detector may flag striking natural features and encourage mistaken interpretations. Combining both approaches could help researchers compare results and reduce dependence on any single assumption about what artificial structures should look like.
An Initial Test—and Its Limits
The paper describes an early self-supervised method developed for lunar imagery. It identified the Apollo 17 lander, a known human artifact, along with natural features of scientific interest, including recent impacts and volcanic pits. That result shows the method can flag unusual objects. It does not provide evidence of extraterrestrial technology: the lander is human-made, and the natural findings demonstrate why anomalies require careful interpretation.
The authors stress that the field remains at an early stage. No candidate extraterrestrial surface technosignature or confirmed alien artifact has been reported, and no quantitative upper limits have been published. The paper also does not estimate the probability that such artifacts exist. Searches could produce false positives by misidentifying natural features, or false negatives by missing artifacts that do not match a model’s expectations.
A Practical Addition to the Search for Life
Despite those uncertainties, the authors argue that surface searches are a promising, relatively low-cost addition to technosignature research. Existing archives make it possible to begin without waiting for new missions. The work could also benefit planetary science by drawing attention to unusual natural formations that merit closer study.
The next steps are to run supervised and unsupervised searches side by side, check promising candidates against independent datasets, and pursue in situ investigation where possible. The central proposal is modest but testable: look for persistent, non-random changes to planetary surfaces, and measure what the available evidence can—and cannot—rule out.
