An AI algorithm says domestic security risk is down

Mass shootings are up in the U.S., but domestic security risks have been falling for a month, according to an artificial intelligence algorithm used by GeoQuant, a Berkeley-based political risk firm.

Meanwhile, the multiple crises abroad — North Korea, Syria and Russia among them — have raised U.S. geopolitical risk, said Mark Rosenberg, GeoQuant's CEO. In late May, the firm's measure of domestic security risk crossed over the geopolitical risk line.

What it means: Unlike on Wall Street, where algorithms are now widely used for trading and forecasting, machine learning tools are new in political risk. GeoQuant is one of the first firms testing them in the field, developing forecasts alongside political scientists. According to its current forecast, the threat of U.S. political violence remains high, with the decline in security risk coinciding with the Trump administration's greater openness to police crackdowns, along with the fall of border crossings and immigration.

Data: GeoQuant; Chart: Lazaro Gamio / Axios