Skip Navigation
Analysis

How to Reform Police Monitoring of Social Media

Reforms should place guardrails on law enforcement access to social media and codify legal recourse for impact.

Civil rights protest­ers have long been the target of state surveil­lance. Recent efforts to keep tabs on Black Lives Matter protests and demon­stra­tions against family separ­a­tion, for instance, echo the FBI’s counter-intel­li­gence program used to spy on civil rights lead­ers from Martin Luther King, Jr. to Huey Newton

Yet today’s surveil­lance analysts have a new source of inform­a­tion: social media. Consider the recent protests over the police killings of George Floyd and Breonna Taylor. As demon­stra­tions spread across the coun­try, the FBI and local police monitored social media and made arrests based on what people have posted online. In one instance, police in Wichita, Kansas even arres­ted a teen­ager on suspi­cion of incite­ment to riot­ing based on a threat­en­ing Snapchat screen­shot he shared. The teen’s post added a note caution­ing read­ers to “stay tf away from” his homet­own—rhet­oric inten­ded to denounce the call to viol­ence, not to foment it.

Unfor­tu­nately, the surveil­lance of social media is a grow­ing trend. In recent years, social media posts have landed indi­vidu­als of color in over­broad and unre­li­able gang data­bases, and even been used to justify keep­ing them imprisoned. One New York teen spent more than a year on Rikers Island, based in large part on the district attor­ney’s incor­rect assess­ment that he was a member of a crim­inal gang. The D.A. relied on Face­book photos of the teen with members of a local crew—a group of kids loosely affil­i­ated by block or hous­ing devel­op­ment—and several posts from crew members that he had “liked.” In real­ity, the teen was simply connec­ted to crew members because they were his neigh­bors and family members. The extent to which communit­ies of color are viewed with suspi­cion can create a self-fulfilling proph­ecy where basic social media etiquette is mistaken for member­ship in a crim­inal enter­prise. Social media is highly contex­tual and prone to misin­ter­pret­a­tion, magni­fy­ing the risk that one person’s innoc­u­ous post will be taken as some­thing more sinis­ter.

From protests to public hous­ing, social media monit­or­ing raises civil liber­ties and civil rights concerns that are currently going unad­dressed. Estab­lish­ing a frame­work that balances public safety and the right to privacy, free expres­sion, and equal protec­tion under the law requires updates to our exist­ing regu­lat­ory controls.

How do police monitor social media?

To begin with, how do the police watch social media? Most commonly, an officer views publicly avail­able posts by search­ing for an indi­vidual, group, hashtag, or another search vector. Depend­ing on the plat­form and the search, it may yield all of the content respons­ive to the query or only a portion. When seek­ing access to more than is publicly avail­able, police may use an inform­ant (such as a friend of the target) or create an under­cover account by posing as a fellow activ­ist or allur­ing stranger. This allows officers to commu­nic­ate directly with the target and see content posted by both the target and their contacts that might other­wise be inac­cess­ible to the public.

Police have also used soft­ware to monitor people, groups, asso­ci­ations, or loca­tions in a more auto­mated manner. This soft­ware included tools that mapped clusters of activ­ity and a plat­form for link­ing under­cover accounts. This tactic is less common now after the major plat­forms prohib­ited app developers from receiv­ing auto­mated access to public content for surveil­lance.

Dataminr, the prom­in­ent social media analyt­ics firm, appears to have found a partial work­around to this prohib­i­tion by provid­ing police with “public sector alerts.” Datamin­r’s auto­mated systems analyze public data feeds and deliver auto­mated alerts to law enforce­ment clients, includ­ing the FBI, about shoot­ings and natural disasters. It is unclear whether law enforce­ment can custom­ize the service to go beyond the types of alerts included in Datamin­r’s market­ing mater­ial.

Law enforce­ment may also request social media data as part of a crim­inal invest­ig­a­tion. By deploy­ing subpoenas and warrants, law enforce­ment can collect an array of data directly from social media compan­ies. For example, a template warrant drawn up by the Depart­ment of Justice to serve on Face­book contem­plates collect­ing an array of data, includ­ing contact inform­a­tion, photos, status updates, private messages, friends lists, group affil­i­ations, “friend” requests, future and past event post­ings, privacy settings, and more.

Social media monit­or­ing is pervas­ive. Accord­ing to a 2017 survey by the Inter­na­tional Asso­ci­ation of Chiefs of Police, 70% of respond­ing police depart­ments use social media for intel­li­gence gath­er­ing and to monitor public senti­ment. Simil­arly, Face­book’s latest trans­par­ency report states that the company received over 50,000 govern­ment requests for data between July and Decem­ber 2019.

What does the law say?  

Despite wide­spread use of social media by police, there are few laws that specific­ally constrain law enforce­ment’s abil­ity to engage in social media monit­or­ing. In the absence of legis­la­tion, the strongest controls over this surveil­lance tactic are often police depart­ments’ indi­vidual social media policies and plat­form restric­tions, such as Face­book’s real name policy and Twit­ter’s prohib­i­tion against using its API for surveil­lance. While the consti­tu­tional land­scape is unsettled, consti­tu­tional protec­tions for privacy, free­dom of speech and asso­ci­ation, and equal protec­tion may provide mech­an­isms for indi­vidu­als to chal­lenge the govern­ment’s collec­tion and use of their social media data.

Fourth Amend­ment protec­tions

The Fourth Amend­ment guar­an­tees the right of the people to be free from unreas­on­able searches and seizures; the inquiry for whether a search was unreas­on­able, outside of the core protec­tions of the Fourth Amend­ment for “persons, houses, papers, and effects,”  gener­ally comes down to whether a person has a “reas­on­able expect­a­tion of privacy” and whether soci­ety recog­nizes that expect­a­tion as reas­on­able. While courts have begun to recog­nize that privacy does­n’t require abso­lute secrecy, they have never­the­less typic­ally held that indi­vidu­als do not have a recog­nized expect­a­tion of privacy in data publicly shared online. As one appeals court put it, “If you post a tweet, just like you scream it out the window, there is no reas­on­able expect­a­tion of privacy.”

When police want access to inform­a­tion that is not read­ily avail­able to the public, however, the Supreme Court gener­ally requires law enforce­ment to meet a higher stand­ard. The Court has held, for instance, that when law enforce­ment uses digital tools to view inform­a­tion that would other­wise require a warrant to obtain, the use of that tool requires a warrant as well. Never­the­less, courts have gener­ally allowed police to engage in under­cover oper­a­tions (both online and in the real world) without obtain­ing a warrant—­though indi­vidual law enforce­ment agen­cies may put addi­tional restric­tions in place.

This permissive approach largely grows out of a separ­ate legal prin­ciple called the “third-party doctrine.” Under this doctrine, when people share inform­a­tion with a third party, whether another person or a busi­ness, they should expect that the data could be disclosed to the govern­ment.

Courts have exten­ded this reas­on­ing to social media. In United States v. Mere­gildo (2012), for instance, a New York district court found that while a person’s privacy settings might have signaled an intent to main­tain secrecy, any expect­a­tion of privacy was “extin­guished” when they shared content with Face­book friends because those indi­vidu­als were free to use that inform­a­tion however they saw fit—in­clud­ing by shar­ing it with law enforce­ment.

But the bound­ar­ies of the third-party doctrine are becom­ing murkier, and the Supreme Court is increas­ingly signal­ing a need to recon­sider the doctrine in light of tech­no­lo­gical advances. Most recently, in the 2018 case Carpenter v. United States, the court ruled that despite the third-party doctrine, the police had to obtain a warrant before it could access histor­ical loca­tion inform­a­tion held by cell phone providers. And in the Supreme Court’s 2012 decision in United States v. Jones, which held that the govern­ment had to get a warrant to install a GPS device that enabled constant loca­tion track­ing, Justice Sonia Soto­mayor observed in her concur­rence that secrecy might need to stop being treated as a “prerequis­ite for privacy” in order to account for the volumes of sens­it­ive data shared with third parties on a daily basis. 

The applic­a­tion of the Fourth Amend­ment has been rooted in prac­tical limit­a­tions, but the force multi­plier effect of social media surveil­lance calls those limit­a­tions into ques­tion. If police want to phys­ic­ally trail a car, there are limits to the number of vehicles that can be followed. Having to make determ­in­a­tions about how best to alloc­ate an officer’s valu­able time provides some rough limits, even if those resource decisions are not always made judi­ciously. By contrast, a single officer can monitor the social media accounts of dozens of people all at once, without having to leave her work­sta­tion. Under­cover social media accounts offer greatly expan­ded power as well. An officer who would have to take care to create and main­tain a single under­cover persona in real life could have twenty differ­ent perso­nas online. And soph­ist­ic­ated soft­ware may be able to assist police in monit­or­ing thou­sands of accounts at the push of a button.

In 1983, the Supreme Court predicted that this kind of wide-scale surveil­lance could trig­ger a differ­ent legal analysis. “If such drag­net-type law enforce­ment prac­tices…should even­tu­ally occur,” it may be time “to determ­ine whether differ­ent consti­tu­tional prin­ciples may be applic­able,” Justice William Rehnquist wrote in United States v. Knotts, in which the court decided that track­ing a suspect using a surrepti­tiously planted beeper device did not consti­tute a Fourth Amend­ment search.

Finally, law enforce­ment can also seek account inform­a­tion directly from social media compan­ies. Under the Stored Commu­nic­a­tions Act (SCA), law enforce­ment can serve a warrant or subpoena on a social media company to get access to inform­a­tion about a person’s social media profile. The SCA also permits service providers to volun­tar­ily share user data without any legal process if delays in provid­ing the inform­a­tion may lead to death or seri­ous injury.

While numer­ous defend­ants have chal­lenged the consti­tu­tional valid­ity of warrants to search their social media data, courts have gener­ally upheld the warrants. For example, courts have upheld warrants look­ing for IP logs to estab­lish a suspect’s loca­tion, for evid­ence of commu­nic­a­tions between suspects, and to estab­lish a connec­tion between co-conspir­at­ors.

First and Four­teenth Amend­ment protec­tions

The First and Four­teenth Amend­ments offer protec­tions where surveil­lance is based on polit­ical or reli­gious beliefs, asso­ci­ations, racial and ethnic iden­tit­ies, and other protec­ted categor­ies or activ­it­ies. Surveil­lance can have a chilling effect on First Amend­ment freedoms, and surveil­lance that dispro­por­tion­ately targets a protec­ted class can give rise to equal protec­tion harms under the Four­teenth Amend­ment.

Social media monit­or­ing can have seri­ous chilling effects on protec­ted speech and asso­ci­ation. When police target indi­vidu­als for surveil­lance because of their polit­ical view­points, people may choose to censor their online activ­ity and asso­ci­ations to reduce the risk of govern­mental monit­or­ing. Like­wise, law enforce­ment may use social media to compile dossiers on people on the basis of First Amend­ment-protec­ted activ­it­ies and may share them among local, state, and federal agen­cies—this can increase the risk that protest­ers later face retali­at­ory target­ing in the form of deport­a­tion proceed­ings or unre­lated crim­inal prosec­u­tions.

The case law on whether these impacts give rise to concrete First Amend­ment claims contin­ues to evolve. In Laird v. Tatum (1972), the Supreme Court adop­ted a restrict­ive approach to the First Amend­ment, ruling that plaintiffs in the case did not have stand­ing to chal­lenge an Army intel­li­gence program stood up in response to 1960s civil rights protests. That program collec­ted inform­a­tion from news­pa­pers and police depart­ments and sent Army intel­li­gence agents to attend public meet­ings. But, the Court ruled, the mere exist­ence of a surveil­lance program that may have chilled speech did not cause suffi­cient harm to give the plaintiffs the right to sue under the First Amend­ment.

Where the target­ing of First Amend­ment-protec­ted activ­ity leads to a concrete harm, affected indi­vidu­als may have stand­ing to raise a claim. For example, in Baird v. State Bar of Arizona, a lawyer was preven­ted from join­ing the state bar because she refused to answer if she had ever been a member of the Commun­ist Party. There, the Supreme Court ruled that the govern­ment “may not inquire about a man’s views or asso­ci­ations solely for the purpose of with­hold­ing a right or bene­fit because of what he believes.” Apply­ing this ruling to the digital era, if a person’s social media is surveilled on the basis of her polit­ical beliefs or asso­ci­ations, and she is later denied a civil bene­fit or prosec­uted for an unre­lated crime in retali­ation for the beliefs or asso­ci­ations revealed, she should have stand­ing to bring a First Amend­ment claim. In a time when social media monit­or­ing allows the govern­ment to discover a person’s beliefs and asso­ci­ations easily and covertly, there is a partic­u­lar need to ensure that this inform­a­tion is not misused to target disfavored indi­vidu­als for adverse govern­ment action. Unchecked, discrim­in­at­ory surveil­lance can have chilling effects on unpop­u­lar but lawful asso­ci­ations.

Fortu­nately, in Hassan v. City of New York (2015), the Third Circuit ruled that where discrim­in­at­ory govern­ment monit­or­ing dissuades indi­vidu­als from exer­cising their consti­tu­tional rights, they may have stand­ing to chal­lenge the surveil­lance. The Hassan court eval­u­ated the NYPD’s post-9/11 surveil­lance of Muslim communit­ies in New York and New Jersey. This surveil­lance program targeted mosques, student groups, busi­nesses, cafes, and more, lead­ing many indi­vidu­als to limit or with­draw from being active parti­cipants in their community. The Third Circuit ruled that where surveil­lance is racially or reli­giously biased, or under­taken in retali­ation for exer­cising First Amend­ment rights, impacted indi­vidu­als have stand­ing to chal­lenge the prac­tice in court—even if the discrim­in­a­tion is not “motiv­ated by ill will, enmity, or hostil­ity.”

Extend­ing Hassan to social media monit­or­ing, surveil­lance that targets protec­ted speech or dispro­por­tion­ately targets a racial or reli­gious group, and leads to a concrete harm, can give rise to a viable First or Four­teenth Amend­ment chal­lenge, even if the surveil­lance was not anim­ated by “overt hostil­ity or preju­dice.” By inten­tion­ally target­ing Black Lives Matter activ­ists, police may be enga­ging in this kind of discrim­in­at­ory surveil­lance. The same analysis could also apply to the surveil­lance of teens and pre-teens of color suspec­ted of gang activ­ity.

These discrim­in­at­ory surveil­lance programs, which target their online lives for dispar­ate treat­ment based on their ethni­city, may stamp them with a badge of inferi­or­ity that viol­ates their rights to equal protec­tion under the law. In Boston, a defense attor­ney recently won a discov­ery order requir­ing the police depart­ment to produce records related to its surveil­lance on Snapchat, which will offer an oppor­tun­ity to determ­ine whether the program dispro­por­tion­ately targets Black and Latino men.

Room for reform

The avail­ab­il­ity of social media has dramat­ic­ally expan­ded the scope of law enforce­ment surveil­lance. At the same time, few depart­ments have publicly avail­able policies govern­ing their use of social media for intel­li­gence, data collec­tion, and crim­inal invest­ig­a­tions. There are a number of prac­tical steps that can be taken to begin to address this gap.

  1.  Every juris­dic­tion should be required to hold public hear­ings and obtain local govern­ment approval before police engage in social media monit­or­ing. Where depart­ments are already engaged in this prac­tice, they should pause the bulk of these oper­a­tions pending public hear­ings and eval­u­ate whether exist­ing surveil­lance programs dispro­por­tion­ately target consti­tu­tion­ally protec­ted groups or asso­ci­ations.
  2. Every law enforce­ment agency that uses social media for data gath­er­ing purposes should have a publicly avail­able policy that describes their use of social media. These policies should detail the restric­tions and proced­ures for social media monit­or­ing and should specify the stand­ards govern­ing collec­tion, use, reten­tion, and shar­ing of personal inform­a­tion. Social media policies should contain clear prohib­i­tions against surveil­lance based on race, reli­gion, gender, sexual orient­a­tion, immig­ra­tion status, or a person’s exer­cise of First Amend­ment freedoms. The policy should also specify the legal processes that must be followed before law enforce­ment can seek social media data from compan­ies. The restric­tions imposed by a depart­ment’s social media policy should be legally enforce­able, such as by state attor­neys general or the Depart­ment of Justice.
  3. Not every invest­ig­a­tion warrants the use of invas­ive covert accounts. There should be strict controls on the use of this tech­nique, includ­ing ongo­ing monit­or­ing, super­vis­ory approval and over­sight, and time limit­a­tions. Law enforce­ment should be banned from imper­son­at­ing an actual person without that person’s permis­sion. Where law enforce­ment wants to use a covert account, they should be required to docu­ment that no less-invas­ive means are avail­able and to submit the docu­ment­a­tion to an external body for over­sight and approval. Judi­cial over­sight of online under­cover activ­ity would offer the most robust protec­tion. 
  4. Police are preven­ted from inter­view­ing minors without noti­fy­ing their parent or guard­ian. This protec­tion should be exten­ded to the online space with a flat prohib­i­tion against police connect­ing with minors via social media.
  5. Social media monit­or­ing should be subject to ongo­ing report­ing and audit require­ments. For example, police should have to regu­larly disclose inform­a­tion such as the number of social media invest­ig­a­tions that are open and closed and those that are exten­ded past their original clos­ure date. The reports should also indic­ate where invest­ig­a­tions may impact protec­ted classes of people. Each police depart­ment’s social media monit­or­ing prac­tices should be audited by an inde­pend­ent entity on an ongo­ing basis to ensure compli­ance with consti­tu­tional protec­tions and safe­guards.

These recom­mend­a­tions are inten­ded as start­ing points for enact­ing needed reforms. They repres­ent a regu­lat­ory floor, and juris­dic­tions should be empowered to enact stronger prohib­i­tions against certain types of monit­or­ing. This may prove neces­sary in juris­dic­tions where police depart­ments exhibit repeated and consist­ent prac­tices that viol­ate consti­tu­tional rights and freedoms. Estab­lish­ing the right regu­lat­ory controls will require input and action from a number of stake­hold­ers, from communit­ies to civil soci­ety to police depart­ments and local govern­ment to Congress and the Depart­ment of Justice. But this import­ant regu­lat­ory action is over­due and should not be put off any longer.

This piece was origin­ally published by the Brook­ings Insti­tu­tion.