The study’s claim that an AI washer could clean a 5-kilogram load using just 4.8 litres of fresh water has been withdrawn after a review found that test equipment sent much of the rinse water back into the machine. The audit, published on Friday by the Bologna-based Centre for Household Water Research, says the published figure measured water entering the loop, not the total water used.
The result, announced in May as a record low for a full-size household washer, helped Aurelia Market promote five AI-powered models as unusually water-efficient. A fresh test by the centre found that the model at the centre of the paper used 13.9 litres per cycle when measured without recirculation.
How the record was measured
The centre’s audit says a return pipe carried rinse water from the machine’s outlet back to its inlet during 11 of the 14 trials. A meter on the supply line recorded only the fresh water needed to top up the loop; the paper’s methods section did not mention the pipe.
Emilie Králík, a former calibration technician at the centre, said she raised the issue after seeing the record result. “We were measuring how much new water the loop needed, then calling it the machine’s total,” she said. The audit found no evidence that the published readings were altered, but said the test design made the headline comparison invalid.
The record was a measurement of the plumbing around the washer, not the washer alone.
Chiamaka Ndukwu, environmental chemist
The review also found that Aurelia Market supplied €240,000 in equipment and testing support over two years, but the funding was absent from the paper’s disclosures. The centre said the support had been listed in a separate grant register and should have appeared in the journal article. Aurelia Market said it had no role in choosing the protocol or reviewing the results.
Repeated tests, wider fallout
The centre has withdrawn its record claim and asked the journal to correct the paper. Director Ilmars Lielgalvis said the lab would repeat the full comparison on six models, with outside observers present. “We failed to make the test conditions clear, and the result should not have been presented as a measure of ordinary household use,” he said.
Two repeat trials of the same washer at the centre measured 13.9 and 14.2 litres per load after the return pipe was disconnected. That is still below the 17.6-litre average reported for a conventional model in the original study, but the audit says the comparison needs to be rerun under a single, published protocol before any ranking can be trusted.
The disclosure has put pressure on the centre, which advises regional water authorities and publishes guidance used by consumer groups. Its board has paused new industry-funded testing until it adopts a public register of grants and requires researchers to publish plumbing diagrams and raw meter readings alongside results.
Aurelia Market said it had removed the 4.8-litre claim from its product pages and would contact customers who bought the tested model after seeing the promotion. The centre said it would release its revised findings by December; Italy’s national accreditation office has also requested the audit’s test records.
