Silicon Valley

Twitter now has more than 270 million users who actively log in and tweet.

But, according to the latest figures, hidden among these active users are approximately 23 million bots, which tweet automatically.

Once set up, these accounts have little or no human input and range from parody to industry accounts.

For example, @big_ben_clock tweets the word ‘bong’ every hour, on the hour, depending on what the time is.

Elsewhere, @MarmiteBot automatically tweets ‘My mate Marmite’ every time someone mentions the spread on Twitter - even if they’re not tweeting the bot directly.

Similarly there are spelling checker bots, Alan Partridge-inspired automatic replies, and a bot that tweets as ‘the universe’.

Identified spam accounts are not included in this active monthly users figure, but unidentified accounts may be.

In its shareholder filings, Twitter said: ‘In the three months ended June 30, 2014, approximately 11 per cent of all active users solely used third-party applications to access Twitter

However, only up to approximately 8.5 per cent of all active users used third party applications that may have automatically contacted our servers for regular updates without any discernable additional user-initiated action.’ This is said to be lower than expected, however.

Twitter classes spam as any user that creates multiple accounts, either manually or using automated tools.

People who post repeatedly to trending topics to try to grab attention, repeatedly post duplicate updates, and post links with unrelated tweets also violate Twitter’s rules.

It became such a problem on the site, that in 2012 Twitter filed a suit in federal court in San Francisco against five of the most aggressive tool providers and spammers. It then launched tools to help users block spam accounts by flagging up suspicious activity to Twitter.