O Whatsapp banned more than 400,000 accounts from Brazil during Elections 2018 for violating the terms of service. The company, which belongs to Facebook, revealed this information as required by Fake News CPMI. It uses machine learning to detect automated behavior and prevent botting.
Between August 15 and October 28, 2018, WhatsApp banned “more than 400,000 Brazilian accounts for violating our Terms of Service,” says the company in a document sent to the Senate. This period goes from the beginning of the electoral propaganda until the second round of the elections.
As WhatsApp works with encrypted messages, the company uses other signals to find out if users are violating its terms of service: “our decisions against automated activities and mass messaging are based on the behavior of the accounts, rather than the content of messages ”.
This is done in three main ways:
- WhatsApp confirms that the user has access to the line when registering, using the code sent by SMS, and looks for evidence of automated creation en masse from suspicious networks;
- the company looks for signs of abnormal behavior or automation in sending messages: for example, if a number sends messages to others who do not have it in their contact list, this may indicate that the sender obtained the number without the users’ consent;
- the app has options to block and report accounts, and this negative feedback is taken into consideration.
WhatsApp delivered data to Fake News CPMI
Fake News’ CPMI asked WhatsApp to deliver data on accounts banned for suspected bot usage, including “unencrypted metadata, including phone number and profile names.”
WhatsApp says, however, that it no longer has this information for most users. The company follows the Marco Civil da Internet and stores six months of access records, with IP address, date and time – but the elections took place more than a year ago.
For this reason, the company provided only the information that needed to be preserved on accounts that “were the subject of data disclosure to Electoral Courts in Brazil”. They were banned during the election period on suspicion of spam and mass messaging. Data on these accounts have been sent to CPMI, but are not publicly available.
The Fake News CPMI was created in July and aims to identify accounts and profiles that have disseminated fake news during the 2018 elections.
With information: federal Senate.