Social Media Monitoring Continues to Concern Civil Rights Groups

Using machines to translate text and its meaning has been a point of contention among rights groups, social media companies and law enforcement.

by Marissa Lang, San Francisco Chronicle / October 26, 2017
Shutterstock

(TNS) -- All it took to land a Palestinian construction worker in police custody for several hours on suspicion of planning a terrorist attack last week was a Facebook post — and a message mistranslated by a machine.

This incident, which garnered international attention Tuesday when Facebook publicly apologized for the mistaken translation, sums up why U.S. civil rights attorneys and Internet privacy groups are disturbed that the U.S. government has doubled down on efforts to gather data from social media accounts of all incoming immigrants.

Civil rights advocates say using machines to translate messages and interpret context or meaning in social media posts is rife with problems.

One mistake can result in wrongful denials of entry, arrests and worse, immigration lawyers said, while retaining sensitive information on people’s social media habits may endanger immigrants in their home countries and threaten the free expression of non-U.S. citizens and citizens alike.

“Let’s say someone is applying for a visa from a country with minimal human rights. They’re not going to want to use their real name in their online speech or activism,” said Christina Sinha, staff attorney and program manager of the national security and civil rights program at the Asian Law Caucus in San Francisco. “The government is forcing people to link their anonymous online identities to traceable, shareable information that puts people and their families at risk for no real reason. There’s no evidence that collecting and holding onto this information makes any of us any safer.”

The Department of Homeland Security last week issued a notice in the Federal Register that confirmed the federal government has been collecting “social media handles, aliases, associated identifiable information and search results” on immigrants — including permanent U.S. residents and naturalized citizens — since 2012, a program that was expanded in 2015.

A new federal rule that took effect last week may lead to an increase in both scope and frequency of this data collection, lawyers said. It also revealed for the first time that the government intends to store that information indefinitely.

It’s the latest move in an ongoing push to monitor people’s online activity in the name of national security. Several civil rights and Internet advocacy organizations including the American Civil Liberties Union and the Electronic Frontier Foundation have decried the government’s use and storage of people’s digital data.

Homeland Security and Trump administration officials have defended the efforts as a means of preventing terrorism.

“DHS, in its law-enforcement and immigration-process capacity, has and continues to monitor publicly available social media to protect the homeland,” department spokeswoman Joanne Talbot said in a statement.

But several experts and federal officials remain unconvinced that this kind of monitoring makes the U.S. any safer.

The DHS inspector general published a report in February that said the effectiveness of these social-media screening programs remains unclear at best.

Part of that, the report said, is that federal agencies haven’t specified what they’re looking for.

Though most people’s social media accounts are filled with seemingly inconsequential updates about daily life, en masse, civil rights advocates said, that information can be used to create detailed maps of where people travel, who they know, what organizations they follow and more.

“A lot of people think there can’t be any real privacy interest in this information if it’s already online, but that way of thinking misses a lot of what is actually happening here,” said Hugh Handeyside, a senior staff attorney with the ACLU’s national security team. “When we post something on our Twitter or our Facebook feeds, we might expect that people will see it, but we don’t expect law enforcement agencies will basically create digital dossiers about us that can be incredibly revealing.”

It may, civil rights attorneys worry, even lead to a situation like the one that transpired in Israel last week.

The construction worker, who has not been identified in media reports from Israel, posted a photo of himself leaning nonchalantly against a bulldozer near the West Bank settlement of Beitar Illit, near Jerusalem. In his left hand, he balanced a cup of coffee and a lit cigarette. Underneath the photo were Arabic words that roughly translate to “good morning.”

But a Facebook algorithm used to translate one language to another converted the words to something very different. In Hebrew, the algorithm translated the message to “attack them.” In English, “hurt them.”

Israeli police officers arrested and detained the man for several hours after his post, according to Israeli newspaper Haaretz.

They asked him about the bulldozer. Was he planning to use it in an attack? Did he want to hurt people?

At no point before the man’s arrest did Israeli authorities consult with an officer fluent in Arabic, according to local reports.

“That’s a prime example of what happens when law enforcement over-relies on machines — even a simple ‘good morning’ got a man landed in custody,” Sinha said. “Even if someone is later found to be innocent, as the man was in this case, that doesn’t mean that their name is scrubbed out of the system. The United States shares information with so many countries ... having a detention on your record could lead to other nightmares if the person travels to other countries that receive their information from the U.S. government.”

The list of entities with which the federal government has reserved the right to share social media information is a long one: state governments, local law enforcement agencies, tribal governments, foreign governments, employers, private companies and more.

Asking for pseudonyms and social media handles that may be used anonymously, lawyers said, could impact the ability of individuals to speak out against oppressive government regimes or engage in online activity that could put them, or their families, in danger.

©2017 the San Francisco Chronicle Distributed by Tribune Content Agency, LLC.

NEW ON THE PODCAST

Tech Startups Can Help Revitalize Both Ends of Your Ballot