Ghost sources: why fake citations are slipping into science

Scientias.nl (Dutch) reports on a PNAS study led by Maxim Topaz of Columbia University that checked about half a million papers in physics, chemistry, engineering and computing. It found 1,135 references that could not be traced to any real source, spread over 619 peer-reviewed articles published between January 2023 and March 2026. The share of affected papers rose from one in 1,600 in 2023 to one in 435 in early 2026. Only two of the 619 have been retracted.
Possible causes include paper mills, carelessness and chatbots that invent sources, though the study did not test where the errors came from. For freethinkers, who often say 'show me the evidence', it is a reminder to check that the evidence exists at all.
Source: Daniel van Veelen, Scientias.nl (Dutch) (2026-11-10)
https://scientias.nl/spookbronnen-rukken-op-aantal-nepverwijzingen-in-de-wetenschap-ruim-verdrievoudigd/
Summarised by the FTI newsdesk.