French government uses AI to Discover undeclared swimming pools and tax them

The French government has collected nearly €10 million in additional taxes after using machine learning to spot undeclared swimming pools in aerial photos. In France, housing taxes are calculated based on a property’s rental value, so homeowners who don’t declare swimming pools are potentially avoiding hundreds of euros in additional payments.

The attempt to find the hidden pools started last October, when Google and IT company Capgemini analyzed aerial pictures taken by France’s National Institute of Geographic and Forest Information that were made available to the public. Pools were identified using software, and the data was then compared with national tax and property databases.

The study has only examined images from nine of France’s 96 metropolitan departments, placing it within a rather constrained scope. According to a statement released this week by France’s tax administration, the General Directorate of Public Finance (DGFiP), which was originally reported by Le Parisien, officials nevertheless found 20,356 unregistered pools in these locations.

France has an estimated 3.2 million private swimming pools as of 2020, although reports suggest the number has increased as more individuals worked from home during COVID-19 lockdowns and summer temperatures have risen across Europe.

This year, France experienced a severe drought that dried out the country’s rivers, making private pool ownership somewhat divisive. A French Green Party (Europe Écologie les Verts) lawmaker made news when he wouldn’t rule out a ban on the development of new private swimming pools. Such a prohibition, according to the MP Julien Bayou, may be implemented as a “last resort” measure.

He later clarified his remarks on Twitter, saying: “[T]here are ALREADY restrictions on water use, for washing cars and sometimes for filling swimming pools. The challenge is not to ban swimming pools, it is to guarantee our vital water needs.” 

France’s tax offices, the DGFiP (known more commonly as Le Fisc), says it now plans to expand the use of its AI-pool-spotter to the entirety of metropolitan France (excluding the country’s overseas departments), which could net an additional €40 million in taxes.

Early project reports indicated that the machine learning algorithms frequently confused other architectural features, such as solar panel installations, for swimming pools and had an extremely high mistake rate of 30%. Le Fisc claims to have fixed these issues now and plans to utilize its software detection pools to find other unreported and taxable house changes, such as additions and annexes.

According to Antoine Magnant, the deputy director general of public finances, “We are specifically targeting house extensions like verandas, but we have to be sure that the software can find buildings with a large footprint and not the dog kennel or the children’s playhouse,” according to Le Parisien.