AI reveals far more earthquakes at Italy’s Campi Flegrei, raising concerns about increased unrest

Researchers using artificial intelligence have identified more than four times as many earthquakes beneath Italy’s Campi Flegrei volcanic field as earlier methods detected, a finding that expands understanding of the area’s seismic unrest and prompts fresh scrutiny of local hazards.

The study, which applied AI to seismic data in near real time, increased the recorded number of earthquakes between 2022 and 2025 from roughly 12,000 to more than 54,000. The revised catalogue includes many low‑magnitude events that conventional processing missed, and reveals fault lengths that researchers say make earthquakes in the magnitude 5 range possible.

The Campi Flegrei volcanic field, a caldera complex to the west of Naples, has shown renewed activity in recent months. In May the area experienced a magnitude 4.4 quake, the strongest event there in about 40 years. Hundreds of thousands of people live in the broader area and could be affected if larger earthquakes occur or volcanic activity intensifies.

"These long faults suggest that an earthquake in the magnitude 5 range is not out of the question," said study co‑author Bill Ellsworth, who co‑directs the Stanford Center for Induced and Triggered Seismicity. Ellsworth and colleagues said the AI approach provided more precise earthquake locations and magnitude estimates in near real time, improving the resolution of the seismic picture beneath the caldera.

The newly revealed events include swarms of small tremors and previously unrecognized faulting patterns. Researchers say such details matter for hazard assessment because the geometry and extent of faults influence how seismic energy is released and where shaking may be strongest.

Past seismic episodes at Campi Flegrei have caused damage to infrastructure, cracking road surfaces and disrupting services, and the site is watched closely by volcanologists because it is one of the most closely monitored calderas in the world. The scientific community classifies Campi Flegrei as a large, complex volcanic system with a history of explosive eruptions, and its proximity to dense population centers has long made monitoring a priority.

The study did not conclude that an eruption is imminent. Instead, researchers emphasized that expanded detection changes the baseline of background seismicity and offers a more detailed record for scientists and civil authorities to evaluate. Improved earthquake detection can inform models of subsurface processes, including whether earthquakes are driven by tectonic stress, hydrothermal fluid movement or magmatic intrusions.

Applying machine‑learning techniques to seismic networks has become an increasingly common strategy for expanding earthquake catalogues. Algorithms can recognize waveform patterns associated with small events that are buried in noise or overlooked by automated detectors. By producing rapid, high‑resolution catalogues, the tools can reveal temporal clustering, migration of seismicity and connections between faults that were previously hidden.

Researchers plan to continue refining the AI processing and to integrate the expanded earthquake catalogue with other monitoring data, including ground deformation and gas emissions, to build a fuller picture of the caldera’s state. Authorities responsible for civil protection in the region have historically used multiple data streams to assess risk and issue advisories.

Scientists caution that no single dataset is definitive. A rise in detected seismicity can reflect both true increases in underground activity and improvements in detection capability. Nonetheless, the authors said that the newly mapped fault structures and the frequency of events underscore the importance of sustained monitoring and preparedness for seismic hazards in the Naples area.

The findings illustrate how modern data tools can alter assessments of volcanic and seismic risk by revealing previously unobserved activity. As monitoring systems evolve, officials and researchers will use richer datasets to guide hazard planning and public communication about the potential risks posed by one of Europe’s most closely watched volcanic systems.