Last Updated Dec 15, 2015 7:15 AM EST
Following the Dec. 2 San Bernardino massacre that left 14 people dead and 22 injured, authorities have been trying to piece together the background of husband-and-wife shooters Syed Rizwan Farook and Tashfeen Malik and the digital trail they left online. Authorities say Facebook posts by Malik dating as far back as 2012 -- well before her marriage to Farook -- expressed her support for radical, violent jihad.
CBS News has learned that Malik had private chats on Facebook with her sister where she pledged support for violent jihad and said she wanted to participate. These post began before Malik came to the United States and continued after she arrived.
Yet those warning signs were never seen by U.S. officials who granted Malik a K-1 "fiancée" visa to enter the country, prompting new questions about what the government and tech companies should be doing to prevent such threats from slipping through the cracks.
The Department of Homeland Security says it began three pilot programs in 2014 to examine whether screening social media was consistent with current laws and privacy protections, CBS News correspondent Jeff Pegues reported. A review of the policy is underway.
How feasible is it to evaluate posts on Facebook or Twitter as part of a screening process -- especially if people aren't posting under their real names? It's possible, but complicated. Individual investigations would slow down the visa process, potentially hurting business and tourism. And the tech industry has pushed back against past efforts to enlist it in reporting users' content to the government, citing privacy concerns.
The volume of material at issue is enormous. Just look at your personal Facebook news feed and you can see hundreds of posts that range from the mundane to the controversial or maybe even inflammatory. Now multiply that times 1.01 billion daily active users on the site, roughly 84 percent of them from outside the United States and Canada, according to recent statistics from Facebook, posting in many languages and with varying degrees of seriousness or sarcasm, and the scope of the challenge becomes clear.
Needle in a haystack
While finding the next Malik might be akin to looking for a needle in a haystack, sites like Facebook and Twitter do have measures in place to monitor threatening and hateful speech as well posts that could potentially be linked to acts of terrorism.
"Violent threats and the promotion of terrorism deserve no place on Twitter and our rules make that clear," a Twitter spokesperson wrote to CBS News. "We have teams around the world actively investigating reports of rule violations, and they work with law enforcement entities around the world when appropriate."
Twitter's policies specifically state that "users may not make threats of violence or promote violence, including threatening or promoting terrorism."
Similarly, Facebook states on its website that it removes "content that expresses support for groups that are involved in the violent or criminal behavior" related to "terrorist activity" or "organized criminal activity."
Facebook has managed to block ISIS-related accounts and posts more effectively than other sites, Steve Stalinsky, executive director of Middle East Media Research, told Wired magazine last month: "Of all the companies, they're the leader and the best at removing content."
But policing online activity is tricky, cyberwarfare expert David Gewirtz told CBS News. He explained that a company like Facebook faces legal, technical, and ethical considerations in flagging such content.
"From a technical point of view, analytics, natural language artificial intelligence, and sentiment analysis software have all reached the stage that given a body of information (like Facebook postings), it is reasonable to expect them to be able to identify threatening communication," Gewirtz wrote in an email. "That said, it's far from 100 percent and there would also be false positives. Even so, the 'that doesn't seem right' postings could be flagged."
From the legal point of view, Gewirtz added that there are regulations that "require or compel" companies to participate in criminal investigations with the government.
But ethically, companies do face some murky waters.
"How do you separate a high-vitriol comment in the heat of an election season from legitimate threats?" Gewirtz asked. "And is it right for all of us to be watched and potentially flagged to the government? Would we then use or trust these services?"
Gewirtz suggested that one way of dealing with these concerns would be to adjust "the parameters for who to watch." For example, "recent immigrants from certain nations might be reasonable to include in a scan," he said, "while cranky 'Uncle George' in Alabama has been there for 70 years and there's no reason to scan his discussions."
Further complicating things in a case like Malik's is that she published these posts under an alias account, a U.S. law enforcement source told CBS News.
"Facebook does its best to generate accurate identity information," Gewirtz said of trying to identify people who post under false names. "While this is far from perfect ... the 'friending' system of Facebook does provide a data-based bank of raw material with which such a validation could be performed." (He also pointed out that many people want to protect their identities on social media for legitimate reasons -- perhaps because they have health issues, fear persecution, or were the victim of a stalker.)
"Technology can provide a winnowing function, reducing a haystack into a smaller mound, but that mound would still need to be investigated by skilled humans," Gewirtz added.
Catching up with the 21st century
The fact that Malik's vis application was vetted by five separate government agencies who never checked her social media presence suggests the process needs to catch up with the demands of the 21st century.
"There may not be the right protocols or practices or maybe even polcies that allow for the look at social media ... as part of the vetting process," Juan Zarate, CBS News senior national security analyst, said. "It doesn't make sense, it's obviously counterintuitive. We should be looking at any and all information tied to an individual applicant."
While we may not want to be defined by our Facebook profiles, and people are more complicated than a collection of status updates and "likes," social media could still hold clues worth investigating further.
"Sometimes, social media can be a better representation of who a person is than anything else," Tyler Cohen Wood, a cyber security advisor for Inspired eLearning, told CBS News.
Cohen Wood is a former senior officer and a cyber branch chief for the Defense Intelligence Agency at the Department of Defense. She said that while people may post to Facebook thinking that "no one is looking," the wide extended network of friends on the site means that something like inflammatory speech would be easily discoverable.
Over the summer, tech giants like Twitter, Yahoo. Facebook, and Google pushed back against a Senate proposal that called for the companies to alert federal authorities about suspicions of terrorist activities.
All of these sites have policies banning terrorist threats or content like video of beheadings, but they expressed concern that such legislation could get companies in hot water for missing a post or a tweet -- which are sometimes vague -- that could signal an impending terror attack.
Sen. Dianne Feinstein, D-Calif., reintroduced the bill last week.
Beyond company concerns, users have expressed doubts over whether their own online privacy could be compromised by a more thorough review of social media posts.
"Finding and fighting terrorists is very difficult and we do need all the advantages we can get in that fight. But from a sociological perspective, we need privacy and from a purely practical point of view, if someone thinks you're going to blab everything you post to the government, they're just not going to use your service," Gewirtz said.
Cohen Wood suggested that in some instances, the policing done by the social media-using public can be particularly effective.
"If you're friends with someone and you see something like someone spelling out that they are going to shoot a school, that to me doesn't fall into a gray area," she said. "You see this happening a lot more -- just the general public flirting with these dangerous situations. They notice a post that they consider threatening and they contact authorities themselves."