Twitter has confirmed that up to 23m of its active users may actually be bots, automatically updating statuses with no human user on the other end.

The company disclosed the figure in a filing with the US Securities and Exchange Commission, where it said that “up to approximately 8.5 per cent” of accounts are considered bots “without any discernible additional user-initiated action.” The company listed 271m monthly active users (MAUs) at the end of June.

The new disclosure clarifies an earlier statement that 14 per cent of MAUs access the service outside of the official website and mobile apps by using Twitters API. Twitter confirmed the 14 per cent “included certain users who accessed Twitter through owned and operated applications”, meaning programs like TweetDeck and Twitter for Mac, which are favoured by Twitter users with multiple accounts or high usage.

The bots are not necessarily spam accounts, which according to Twitter make up less that 5 per cent of MAUs, and includes many useful, even essential accounts that Twitter users rely on for information. However, once they are set up, they require little to none human interaction, which has a big impact on Twitters ability to reach potential customers with advertising.

Despite reporting 124 per cent increases in year-on-year revenue for Q2 2014, Twitter has yet to report a profitable quarter since it went public in 2013. The majority of its revenue is generated by advertising, and this latest news could see advertisers lose some of their faith in the social networks ability to deliver real results.