Antarctica was not lacking earthquakes. It was lacking the ability to hear them.
Beneath Antarctica's ancient ice, hundreds of small earthquakes have been speaking for decades in a language too quiet for human instruments to hear — until a machine learning algorithm, trained to listen differently, found more than 500 tremors hidden inside seismic records stretching back to 2001. The discovery, centered deep beneath David Glacier at a boundary where cold East Antarctic crust meets warmer West Antarctic rock, challenges the long-held assumption that the continent is seismically dormant. What emerges is not merely a geological correction, but a reminder that absence of evidence is not evidence of absence — and that the Earth's inner life may be far richer than our tools have allowed us to perceive.
- Over 500 earthquakes, ranging from magnitude 1.6 to 3.5, had been hiding in plain sight within Antarctic seismic records for more than two decades — missed entirely by human analysis.
- The tremors are occurring in a place they shouldn't be: deep inside a tectonic plate, far from any active boundary, where cold rigid crust collides with warmer, softer rock in an abrupt underground transition zone.
- The discovery dismantles a foundational assumption in Antarctic science — the continent was never truly quiet, only unheard, its seismic signals beyond the reach of the tools scientists had been using.
- Researchers are now turning AI loose on existing seismic archives, hoping to extract forecasts about ice loss and sea-level rise from data that has been sitting unread for years.
- The implications extend far beyond Antarctica: if machine learning can find this many hidden quakes on one continent, deep intraplate earthquakes worldwide may be dramatically undercounted, reshaping how scientists understand tectonic stability globally.
Beneath Antarctica's ice, hundreds of earthquakes have been rumbling for years without anyone noticing. It took a machine learning algorithm to find them — more than 500 tremors that seismologists had passed over while scanning decades of records from stations across the frozen continent.
The earthquakes are small, ranging from magnitude 1.6 to 3.5, and they sit between 60 and 90 miles underground beneath David Glacier, a massive outlet glacier that drains roughly four percent of the East Antarctic Ice Sheet into the ocean. Researchers had examined records from 49 stations across two time windows — 2001 to 2004 and 2012 to 2015 — but human analysis had simply missed these signals. The AI did not.
What makes the discovery unusual is where the quakes are occurring. Seismic activity at these depths typically clusters near tectonic plate edges, but these tremors are happening well inside a plate, far from any obvious boundary. Lead researcher Long Ho explains that the earthquakes cluster where cold, rigid East Antarctic crust meets warmer, softer West Antarctic rock — a dramatic contrast in material properties that causes the crust to adjust under stress.
For decades, Antarctica was considered seismically quiet, but as glaciologist Richard Alley observed, the continent wasn't lacking earthquakes — it was lacking the ability to hear them. The tremors themselves pose no threat to the ice sheets above, but understanding the forces beneath the ice matters enormously for predicting future ice loss and sea-level rise, which are critical concerns for coastal communities worldwide.
The broader implication is striking: if machine learning can uncover this many hidden earthquakes in Antarctica, deep earthquakes inside continental interiors may be far more common than scientists have assumed — and as the algorithms improve, the questions about what that means for ice, oceans, and the communities that depend on them will only grow more urgent.
Beneath the ice of Antarctica, hundreds of earthquakes have been rumbling for years without anyone noticing. It took a machine learning algorithm to find them—more than 500 tremors that seismologists had missed while scanning through decades of data from seismic stations scattered across the frozen continent.
The earthquakes themselves are small, ranging from magnitude 1.6 to 3.5, and they sit deep underground, between 60 and 90 miles down in the crust beneath David Glacier, one of the massive outlet glaciers that drains roughly 4 percent of the East Antarctic Ice Sheet directly into the ocean. Researchers had examined seismic records from 49 stations across two separate time windows—2001 to 2004 and 2012 to 2015—but human analysis had simply passed over these signals. The AI did not.
What makes the discovery strange is where the earthquakes are happening. Seismic activity at these depths typically clusters near the edges of tectonic plates, in places where one plate is sliding beneath another or where boundaries are actively grinding. But these Antarctic tremors are occurring well inside a plate, far from any obvious tectonic edge. According to Long Ho, the lead researcher on the study, the earthquakes cluster at a subsurface boundary where two very different rock environments meet. The cold, rigid crust and upper mantle of East Antarctica abuts against warmer, softer rock beneath West Antarctica, and that contrast creates an abrupt shift in how the rock behaves under stress. The earthquakes are essentially the crust adjusting to that dramatic change in material properties.
The finding reshapes what scientists thought they knew about Antarctic seismicity. For decades, the continent was considered relatively quiet seismically—a place where deep earthquakes simply did not happen with any frequency. But that assumption was built on a limitation of the tools available to listen for them. Richard Alley, a glaciologist not involved in the research, put it plainly: Antarctica was not actually lacking earthquakes. It was lacking the ability to hear them. Now, with better algorithms sifting through old data, the picture has changed.
The practical implications ripple outward from the ice. These particular tremors are too small to threaten the ice sheets above them or the Antarctic ecosystem. But understanding the forces at work beneath the ice matters enormously for predicting how Antarctica will behave in the future. The continent's ice plays a central role in global sea-level rise, and accurate forecasts of how much ice will be lost and how quickly are critical for coastal communities preparing for change. Better tools for analyzing seismic data mean scientists can extract discoveries from records that already exist, rather than waiting years for new data to accumulate.
The broader implication is even larger. If machine learning can uncover this many hidden earthquakes in Antarctica, it suggests that deep earthquakes inside continental interiors may be far more common than Earth scientists have realized. That could fundamentally reshape how researchers think about tectonic activity in regions long assumed to be relatively stable. As the algorithms improve, the continent's hidden seismic life will likely become even more visible—and the questions about what it means for ice, for oceans, and for the communities that depend on them will only grow more urgent.
Citas Notables
The earthquakes occur where the cold, rigid crust and upper mantle beneath East Antarctica meets warmer, softer rock beneath West Antarctica, and this contrast creates an abrupt change in tectonic strength.— Long Ho, lead researcher
Antarctica was long considered to largely lack earthquakes. Now, we know that the apparent lack of earthquakes was really a lack of tools to listen to earthquakes.— Richard Alley, glaciologist
La Conversación del Hearth Otra perspectiva de la historia
So these earthquakes were always there, but nobody saw them until now?
Exactly. The data was sitting in seismic station records for years. Human analysts were looking at the same information, but the signals were too faint or too numerous to catch by hand. The machine learning algorithm could process all of it at once and spot patterns humans would miss.
Why does it matter that they're happening inside the plate instead of at the edges?
It challenges what we thought we knew about where earthquakes happen. We've always expected deep earthquakes to occur where plates collide or slide past each other. Finding them in the middle of a plate, at a boundary between two different rock types, suggests the Earth's interior is more active and complex than the old models accounted for.
Are these earthquakes dangerous?
Not these ones. They're too small to damage anything. But they're a window into the forces working beneath the ice, and understanding those forces helps us predict how the ice sheet will respond to warming and how much sea level will rise.
What's the bigger picture here?
If Antarctica has this many hidden earthquakes, other continents probably do too. We may have been systematically underestimating seismic activity in places we thought were quiet. That changes how we think about continental stability and the forces shaping the planet.
Can these tools help with other problems?
Absolutely. The same approach—using AI to reanalyze old data—could unlock discoveries in other fields where we have decades of records but limited tools to interpret them. We're not waiting for new data anymore. We're learning to listen better to what we already have.