Robot journalist causes mass panic after reporting fake earthquake

The only tremors from a reported major earthquake off the California coast came on the internet.

Seismologists said Wednesday's automatically generated report of a magnitude 6.8 quake in the Pacific Ocean 10 miles west of Santa Barbara was a false alarm based on a quake that happened in the same area nearly a century ago.

"The quake did happen, but it happened in 1925," said Rafael Abreu, a geophysicist from the US Geological Survey.

The report caused huge ripples on Twitter, where dozens of automated tweets were generated, and concerned citizens were hoping the people of California were OK.

But suspiciously there were no tweets or posts from anyone having felt the temblor, which usually precede the official alerts and come in big numbers. A quake of that size would have been felt by millions.

It turns out that researchers from the California Institute of Technology had been using new information to relocate the epicenter of a 1925 earthquake in the Santa Barbara Channel, which leveled several buildings and killed 13 people.

Somehow this process set off the automated alert that went out to email accounts.

The fake quake never appeared on the USGS website.

A USGS statement said the research "was misinterpreted by software as a current event. We are working to resolve the issue."

The report also set newsrooms around the country on edge. Reporters and editors often use the automated emails from the USGS to begin their coverage, though the reports themselves warn that they contain automated information that has not been reviewed by a seismologist.

The Los Angeles Times was put in an especially bad spot. The newspaper sent out a robotic story that it quickly had to retract.

"We have an algorithm (Quakebot) that automatically writes stories about earthquakes based on USGS alerts," the Times said in a tweet. "The USGS alert was incorrect."

False alarms through the service are fairly common, but they rarely report quakes so big or in such populated areas.

Often when there is a large earthquake in some part of the world, it sets off reports of minor quakes in distant areas. The USGS usually retracts these quickly.

Oddly, while the quake was from the distant past, the report itself put it in the near future. It listed the date as June 29, the same date as the 1925 quake. But the report said the quake happened in the year 2025.

That degree of advance prediction would truly be a miracle.