Type to search

FinTech Tech

Twitter Partner’s Alerts Highlight Divide Over Surveillance

Share
Twitter Partner’s Alerts Highlight Divide Over Surveillance

Twitter Inc.


TWTR 1.34%

Chief Executive Jack Dorsey has pledged that Twitter’s data won’t be used for government surveillance, a commitment the company affirmed four years ago in responding to allegations of widespread police monitoring of Black Lives Matter protesters.

A Twitter data partner is selling a service to law enforcement and other agencies that monitors Twitter on their behalf and provides alerts that include information about potential demonstrations and possible criminal behavior, documents and interviews show. That has the companies at odds with privacy advocates over what level of social-media monitoring qualifies as surveillance.

In recent months, the partner, Dataminr Inc., provided alerts to police and other government clients that included Twitter handles of users discussing plans for protests or where activists were blocking streets, according to the emailed alerts viewed by The Wall Street Journal. Twitter’s rules prohibit partners from using its data for “tracking, alerting or monitoring sensitive events,” specifically including protests and rallies.

One alert was sent early on June 13 to Department of Defense staffers and said that a protest was planned that afternoon at an address near Centennial Olympic Park in Atlanta in response to the police shooting of a Black man named Rayshard Brooks the night before.

Dataminr alerts sent to the Minneapolis Police Department in May and June called attention to protests as well, including tweets on the whereabouts of “a small group of young protesters” chanting “peaceful protest,” and others providing the location of sit-ins. The Minneapolis alerts were reported earlier by the Intercept.

Mr. Dorsey, during 2018 congressional testimony about Twitter’s refusal to offer data products to intelligence agencies, said: “We have a right and responsibility to protect the privacy of the people on Twitter from constant, 24-7 surveillance.” Twitter’s written policy forbids use of its data “by any entity for surveillance purposes, or in any other way that would be inconsistent with our users’ reasonable expectations of privacy. Period.”

In practice, Twitter defines surveillance as the ongoing monitoring of specific people and organizations, a Twitter executive told the Journal. The executive said it was reasonable for Dataminr to alert law enforcement to information that might have public safety value. “Even peaceful protests can cause disruptions,” the executive said.

Dataminr says First Alert—its product for police, military and other government clients—provides only news alerts based on public tweets, and is designed to help first responders react quickly to breaking events. The company says First Alert was built with Twitter’s help to preclude surveillance by government customers, and doesn’t enable them to search for or track specific accounts, among other limitations. Government clients receive alerts tailored to their interests, though “protests” isn’t a topic they can choose, Dataminr says.

“First Alert provides a public good. And importantly, it does so with maximal protections for privacy and civil liberties,” Dataminr said in a statement. Twitter said Dataminr complies with its surveillance rules.

Many law-enforcement officials and others say that people who post statements and images publicly on social media should expect that police might see those materials, and that such observation doesn’t amount to surveillance.

On the other side of the debate, many civil liberties advocates define social-media surveillance broadly as systematic collection and analysis of information about people, groups and events targeted by authorities. Advocates contacted by the Journal said Dataminr’s broad monitoring of Twitter on behalf of government clients means Twitter is, in effect, enabling constant monitoring of its users.

“Any reasonable definition of surveillance should include trolling around Twitter using special access to listen to what people are talking about,” said Faiza Patel, director of the Liberty and National Security program at the Brennan Center for Justice, a policy think tank and advocacy group that promotes privacy rights among other issues. “Having a third party do it doesn’t change the nature of what’s happening.”

Dataminr, which also has products for businesses and newsrooms, pays for access to the entire fire hose of public tweets under Twitter’s Data Partner program. As the partner serving government clients seeking real-time information about events, Dataminr ingests hundreds of millions of tweets daily, analyzing them with a speed and comprehensiveness that would otherwise be unavailable.

Dataminr has cited the unique value of its filtering in fighting public records requests in Illinois and Minneapolis by the Journal covering its First Alert emails. “While the information itself may be public and theoretically discoverable, the fact that it has been identified by Dataminr constitutes the most highly valuable, proprietary, and extremely sensitive dimension of Dataminr’s algorithmic platform,” a Dataminr attorney said in a July letter to the Illinois State Police seeking to block a Freedom of Information Act request by the Journal.

The Minneapolis alerts also included posts on other platforms, and from a variety of people or political groups. One included a post in late May on the Telegram messaging service from a far-right group called Privacy & Security Goys suggesting that anti-Semites might find it an opportune moment to target synagogues because police and other first responders were tied up with riots.

Dataminr says it draws from more than 10,000 public sources. It doesn’t specify all of them or which is largest, but its special access to Twitter’s fire hose is cited in its marketing materials and in clients’ procurement documents.

Among Dataminr’s clients is Storyful, a unit of

News Corp,

which has considered selling Storyful to Dataminr, according to both sides. News Corp also owns Dow Jones & Co., publisher of the Journal.

Government clients have described broad uses for First Alert in procurement documents. The intelligence unit for the Illinois State Police said in a May contract renewal document that its officers can “use Dataminr for a variety of use cases, including the 24/7 monitoring of crimes,” as well as to detect “future criminal action,” according to a contracting document.

Asked for comment, the Illinois State Police told the Journal it doesn’t use Dataminr for surveillance.

Because tweets are already public and no individual is under surveillance, there are no valid civil liberties concerns about Dataminr’s product, the agency said in a statement. “The value in this technology is early notifications or awareness of public safety situations,” it said.

Several former Dataminr employees said they were concerned that the line between First Alert and social media surveillance could be unclear depending on how clients like the military or police used it. One former Dataminr executive said that Twitter executives used the term surveillance vaguely “to allude to creepy things,” and weren’t precise about what activity it entailed.

Twitter said it permits alerts that flag threats to protect lives, but not ongoing monitoring of crimes. “We see a societal benefit in public Twitter data being used for news alerting, first responder support, and disaster relief,” a spokesman said in an email.

Twitter derived $466 million last year, or 13% of its revenue, from the division that licenses data to companies like New York-based Dataminr, in which it used to own a 5% stake.

In June, Dataminr sent several alerts incorporating tweets about Black Lives Matter protests in Louisiana and Atlanta, including the one about the planned protest on June 13, according to alerts sent to Defense Department users that were viewed by the Journal.

Kermit Ray Thomas said he had no idea his tweet about an Atlanta protest had been flagged to the authorities by Twitter partner Dataminr.



Photo:

Raukeem Thomas

A Defense Department spokesman said its staff use a variety of alert systems to make “safety and travel decisions,” and that those alerts sometimes pointed to public events.

The June 13 alert included the Twitter handle of the woman—whose Twitter page clearly identifies her—who posted about plans for the protest, which was scheduled to happen later that day. The Dataminr alert labels her a “local source” and “eyewitness.”

Another of those alerts described video footage taken by a Twitter user identified as @woke_n_restless showing a small group of activists blocking a street following the police shooting of Mr. Brooks. Kermit Ray Thomas, the person behind the @woke_n_restless Twitter handle, told the Journal he had no idea his tweet had been sent to the authorities.

“Alerts on an intersection being blocked are news alerts, not monitoring protests or surveillance,” Dataminr said.

Civil libertarians’ concerns about Dataminr go back several years. In 2016, Twitter responded to claims from civil liberties advocates about the use of social media by “fusion centers,” hybrid state-and-federal hubs for domestic intelligence, to monitor Black Lives Matter activists. Twitter told the American Civil Liberties Union that Dataminr would stop working with the centers.

“Dataminr will no longer support direct access by fusion centers, and has informed us that any accounts accessible by fusion center email addresses have been notified that their access is now terminated,” a senior Twitter executive wrote to the ACLU then.

However, fusion center staffers still have access to First Alert in instances when Dataminr sells to the state police agencies that oversee them, according to interviews and documents in Illinois and Michigan. The ACLU said that practice violated Twitter’s pledge.

“People in fusion centers shouldn’t be accessing Dataminr, and Twitter shouldn’t be in a business of facilitating that,” said Matt Cagle, one of the ACLU researchers to whom Twitter’s letter was addressed. Mr. Cagle also said more generally that he believed sending alerts to Dataminr’s law enforcement clients with protest-related tweets would violate Twitter’s past surveillance commitments.

Dataminr and Twitter said they don’t consider the arrangements with state police to be violating the pledge to the ACLU because the access isn’t direct and First Alert can’t be used for surveillance.

Write to Jeff Horwitz at Jeff.Horwitz@wsj.com and Parmy Olson at parmy.olson@wsj.com

Copyright ©2020 Dow Jones & Company, Inc. All Rights Reserved. 87990cbe856818d5eddac44c7b1cdeb8

Source link

Tags:

Leave a Comment