from the officers:-are-you-tired-of-adhering-to-your-country's-Constitution? dept
Twitter has cut off another social media "surveillance" company from using its API. To date, the platform has forced third-party Dataminr to cut off connections to the CIA, DHS/law enforcement "fusion centers," and Geofeedia. All of these denials of service were the result of the company's policy against use of its API for surveillance.
Very little of what was being done could truly be considered "surveillance," since Dataminr's access to basically every tweet produced did nothing but cull data from public accounts. What Twitter seemed to have more of a problem with was the marketing tactics of companies like Geofeedia, which insinuated their products were perfectly suited for keeping tabs on First Amendment-protected activity, like protests.
As for the CIA and DHS, Twitter apparently felt these government agencies were far more involved in surveillance than the FBI, which just signed a contract with Dataminr for access to its every-tweet-ever API.
The latest recipient of a Twitter disconnection is Canada-based Media Sonar. Again, the issue here appears to be the language used by the company to market its social media monitoring service.
Media Sonar touts its social media monitoring software and algorithms as ideal tools for police and corporations to aggregate and filter data to improve safety and protect corporate assets.
But a U.S.-based investigation turned up marketing language that ran afoul of Twitter's policies, which state that posts on the popular social network should not be mined for surveillance purposes.
Media Sonar's emails to past clients explicitly stated that the software, which allows officers to comb through publicly available posts on the likes of Twitter and Instagram, could help police search for "criminal activity" and "avoid the warrant process" when flagging people who have come under scrutiny.
I'm sure Media Sonar never expected the contents of these marketing emails to be made public, but that's a risk you take every time you send something out inviting law enforcement to use your product to avoid complying with Canadians' rights. Of course, most of what's viewed by law enforcement with tools like these wouldn't require a warrant to obtain.
The move by Twitter may be seen as noble, but it does very little to curb government agencies' monitoring of publicly-available posts. If Twitter users want to remain off the government's radar, it's on them to take more control of the visibility of their tweets. For most users, this isn't a concern and while some may express dismay at law enforcement's use of their posts against them, there's nothing about this outcome that isn't preventable, even without Twitter's periodic announcements that it's cutting another third party off.
The problem isn't with the use of the API so much as it is the interpretation of obtained data. While hashtags may make it easy to track protests and other activity deeply tied to social media interaction, more nebulous data may show correlations that aren't actually there. Overreliance on monitoring tools could result in a lot of false positives, as Canadian Internet Policy staff lawyer Tamir Israel points out.
Israel said most social media monitoring companies rely on algorithms to parse the vast amount of data and pull out meaningful information for clients. Those algorithms, he cautioned, can be misleading.
Israel said they often analyze posts out of context and are unable to account for slang, cultural norms or other factors that give a post meaning.
He cited a recent example of a British tourist who tweeted about his intention to "destroy the United States" on an upcoming trip. His post raised alarm bells with U.S. security, but the tourist had been trying to express his plans to party while abroad using common British slang.
In addition to these concerns are privacy protections granted by Canadian law, which actually gives publicly-available social media posts more protection than those made by US citizens.
[Citizen Lab's Chris Parsons] said law enforcement and federal agencies must demonstrate a need for mining online data, adding that they cannot look through material indiscriminately.
"Just because I say something on Twitter doesn't mean the RCMP can hoover it up," he said. "There has to be a reason, and they have to be able to articulate it."
This may be why the company stealthily sold its product on its warrant-dodging merits. Allowing a third-party to sort and shape the data may allow Canadian law enforcement agencies to wash their hands of any "indiscriminate" hoovering/searching accusations. In any event, Media Sonar's product has suddenly become a lot less useful, and that's going to keep it from being a heavy hitter in the social media monitoring field.