This paper investigates whether the release of market-relevant news in the form of rumours on Twitter can explain the excess of market volatility previously attributed to private information, speculation, and noise traders. We define a simple theoretical model to show that the systematic information content of such rumours should result in detectable price effects in macro-markets. We then pinpoint the arrival of 63 rumours of forthcoming ECB actions over a 420-day sample of one-minute spot EUR-USD rates, and show that there is a real-time, intraday increase in market volatility. This largely unexplored information set can potentially account for significant amounts of unexplained volatility in macro-markets and, therefore, identify a possible explanation of one of the most prominent puzzles in price discovery research.