Trust in information on the web is being damaged by the huge numbers of people paid by companies to post comments online, say researchers.
Fake posters can "poison" debate and make people unsure about who they can trust, the study suggests.
Some firms have created tens of thousands of fake accounts to flood chat forums and skew debate.
The researchers say there are reliable ways to spot fakes and urge websites to do more to police users.
The researchers from Canada and China say paying people to post comments is an "interesting strategy in business marketing" but it is not a benign activity.
"Paid posters may create a significant negative effect on the online communities, since the information from paid posters is usually not trustworthy," they wrote.
In some cases, rival companies have used competing armies of workers to wage comment wars that confused members of the public looking for unbiased information.
The researchers say the fake comments can overwhelm some users, causing them to find it hard to trust any information found online.
They give the example of a spike in activity on a World of Warcraft chat forum on the Chinese website Baidu.
A thread titled "Junpeng Jia, your mother asked you to go back home for dinner!" received over 300,000 replies over a two day period.
A PR company later claimed it had employed 800 individuals to run 20,000 separate accounts on the site to help maintain interest in the videogame while it was down for maintenance.
While the practice of flooding forums with fake comments is most widespread in China, where such posters are called the Internet Water Army, it is becoming common in other nations too.
The US military is known to use fakes to infiltrate chat forums to gather information about potential terror groups.
Similarly many Facebook pages are plagued by bogus friends and "social bots" that are used to stage debates.
Many marketing firms also seed forums with comments in a bid to create "viral" interest in a company or event.
However, fakes can be spotted by analysing their patterns of activity and the words they use, say the researchers.
Fakes are more likely to start new comment threads, make inane comments rather than add to a debate, and repeat former comments with minor changes, the study suggests.
The researchers say they are refining software tools to help website administrators tackle the "painful" problem.