In smart power grids, a smart meter placed at a consumer-end point reports fine-grained usage information to utility providers. Based on this information, the providers can perform demand prediction and set on-demand pricing. However, this also threatens user privacy, since users' specific activity or behavior patterns can be deduced from the finely granular meter readings. To resolve this issue, the authors design Privatus, a privacy-protection mechanism that uses a rechargeable battery. In Privatus, the meter reading reported to the utility is probabilistically independent of the actual usage at any given time instant.