Inland Revenue has denied using a "web crawler" on the internet to find tax evaders.
The internet industry news website Aardvark said a program called IRD-WebCrawler/0.00 testing had accessed a local auction site.
Web crawlers are programs that scan many websites, usually to gather information for search engines.
Aardvark suggestedthat the taxman was using its crawler to try to identify auction site traders who might not have declared income.
Last November, the department seized a database containing the names and contact details of 80,000 owners of ".nz" internet addresses from Domainz, the company that administers internet domain names.
"It could be that the IRD is going to use that list of domain names and a web-crawler to automatically identify those which are e-commerce enabled or which are engaged in trading products or services," said Aardvark editor Bruce Simpson.
IRD national IT manager Tony Lester said the crawler was used to check that links from the department's internal website were up to date.
"Our intranet contains links to external sites, such as the Institute of Chartered Accountants, WestpacTrust or other people who supply useful information.
"This product just checks these linkages are still working."
Mr Lester was "curious" why the crawler had scanned an auction site.
"It may well be that the web server hosting the auction site was also hosting another site that was linked to by our intranet."
