Marwa Fatafta: So, my name is Marwa and I work as a volunteer/activist with a Palestinian dig­i­tal rights group called 7amleh, which means ​“cam­paign” in Arabic. And we have been doc­u­ment­ing and research­ing into human rights or dig­i­tal rights vio­la­tions that are tak­ing place in Palestine and Israel. And one of the most recent case stud­ies or work that we’re look­ing into is the use of pre­dic­tive polic­ing by Israel, which is rather a sen­si­tive issue giv­en that there isn’t a lot that we know about the subject.

I promise I won’t go into politics—maybe just a lit­tle bit, to explain the con­text in which this pre­dic­tive polic­ing sys­tem has devel­oped and how it’s used and what are the human rights and dig­i­tal rights impli­ca­tions of it.

So it starts in October 2015. There was a surge in vio­lence in Israel, in the occu­pied Palestinian ter­ri­to­ries where there was a new trend of a rise of lone wolf attacks where young Palestinians, and most­ly teenagers, would go and com­mit vio­lent attacks against Israeli cit­i­zens or Israeli secu­ri­ty offi­cers, and that insti­gat­ed anoth­er cycle of vio­lence. These young peo­ple were not affil­i­at­ed with any polit­i­cal group. They were not part of an armed cell. There was no sort of polit­i­cal lead­er­ship behind them, or orga­ni­za­tion. They just took mat­ters in their own hands and spon­ta­neous­ly decid­ed that they want to do something. 

And as a result of this let’s say new trend of vio­lence, Israel—or the gov­ern­ment of Israel—blamed social media com­pa­nies for the incite­ment to vio­lence online. And so they blamed com­pa­nies, and name­ly Facebook, for host­ing and facil­i­tat­ing ter­ror­ism and incite­ment to violence. 

And so in September 2016, some Israeli offi­cials said that there’s a spe­cial agree­ment between Facebook and the gov­ern­ment to mon­i­tor and try to tack­le the online incite­ment, some­thing which Facebook actu­al­ly denied in a forum that we orga­nized at 7amleh in Ramallah in the West Bank. They said that there is no spe­cial agree­ment and that the Facebook stan­dards apply to every­body. And they deal with all kind of requests from gov­ern­ments the same way. So they have to take down con­tent if there’s enough evi­dence that they’re actu­al­ly inciteful.

However, expe­ri­ence speaks dif­fer­ent­ly. Because right after a del­e­ga­tion from Facebook vis­it­ed Israel and met with Israeli offi­cials, a num­ber of accounts belong­ing to Palestinian jour­nal­ists and two media out­lets with mil­lions of fol­low­ers were imme­di­ate­ly sus­pend­ed. So Palestinians and activists got online and start­ed a cam­paign under the hash­tag of #FacebookCensorsPalestine, and after that Facebook apol­o­gized and said that was just a mis­take. And so the accounts were restored.

So, aside from try­ing to pres­sure social media com­pa­nies, the Israeli mil­i­tary intel­li­gence start­ed sweep­ing and min­ing social media accounts to try to look for ear­ly warn­ing signs. And so since October 2015, which is the start of the vio­lence, we have doc­u­ment­ed around 800 cas­es of young Palestinians arrest­ed sim­ply because they post­ed some­thing on Facebook. And they have been put under admin­is­tra­tive deten­tion, which means that you are detained with­out any legal process. So you don’t have a tri­al, there’s no charge, and in most cas­es these peo­ple are detained for four to six months. And when that peri­od is over, it gets renewed again.

And some of the lawyers that fol­low up on these cas­es, they said when there was a charge, the charge was incite­ment to vio­lence through social media. That some cas­es, the evi­dence was not pre­sent­ed because it was a state secret or it was an intel­li­gence secret and there­fore they can­not dis­close that evi­dence. Or, some of the detainees said that they were shown screen­shots of their Facebook posts, and they were inter­ro­gat­ed and charged with incite­ment based on that.

I mean, some of the sto­ries… There are so many exam­ples. I mean there are 800 cas­es, right? But some of the sto­ries, an exam­ple is the sto­ry of Tamara, which is a 14 year-old Palestinian girl from East Jerusalem who was arrest­ed in the mid­dle of the night from her home because she wrote on Facebook some­thing like, ​“Please for­give me.” And that could be direct­ed at I don’t know, her boyfriend, or her fam­i­ly, or some teenage dra­ma stuff. But accord­ing to the Israeli mil­i­tary intel­li­gence, it kind of grabbed their atten­tion because they inter­pret­ed that she is about to go and car­ry out an attack.

Another promi­nent case is the case of the Palestinian poet Dareen Tatour. She was… She’s still, actu­al­ly, under house arrest. Same sto­ry. Her home was ran­sacked by secu­ri­ty forces and she was detained from her house in the mid­dle of the night, because as a poet she post­ed a video in which she express­es her anger at the mur­der of Palestinian chil­dren. And that was enough to get her arrest­ed and again, accused of incite­ment. But obvi­ous­ly it’s a clear case of vio­la­tion of free­dom, of expres­sion online. Now she’s banned from using the Internet, she’s under house arrest, and she’s forced to wear an electronic…I don’t know what you call these…bracelet.

So as I said there are many exam­ples. But the rea­son why I men­tion these two cas­es is to high­light the fact that these arrests are not relat­ed to actu­al ter­ror­ism or vio­lent attacks, or even the plan­ning to do ter­ror­ist attacks, but they are relat­ed to cen­sor­ship and the vio­la­tion of free­dom of expres­sion and also pri­va­cy of Palestinians. And this is espe­cial­ly so if we are look­ing at what’s hap­pen­ing on the oth­er side, right. So at 7amleh we did a study or research to see what are Israelis say­ing. Is this incite­ment to vio­lence online only a phe­nom­e­non on the side of Palestinians?

So what we found is that in 2016, Israelis post­ed racist or provoca­tive posts against Arabs and Palestinians on social media every forty-six sec­onds. And almost 60,000 Israeli Internet users wrote at least one post con­tain­ing either racism or hatred towards Arabs, and most of these were relat­ed or had some explic­it or implic­it calls for death and killings of Arabs. And none of these Internet users were charged or arrest­ed or inves­ti­gat­ed for incite­ment. Which again shows that this charge of incite­ment is rather a very broad def­i­n­i­tion, and it could include any kind of resis­tance to Israeli policies.

Now, what we learned this year through inves­ti­ga­tions done by Israeli media is that behind these 800 arrests, or this mass arrest cam­paign, there’s a pre­dic­tive polic­ing sys­tem in place which does social media analy­sis and flags cer­tain peo­ple to be arrest­ed lat­er, based on an attack that may or may not [be com­mit­ted] in the future. In oth­er words these 800—or let’s say the major­i­ty of these cases—were rest­ed based on a machine hunch.

And now, how does this sys­tem exact­ly work? We don’t know the exact inner work­ings of it. It’s a work of mil­i­tary intel­li­gence, after all. But what we know is that… So, when it comes to pre­dic­tive polic­ing there are two kinds of tools. One that is location-based, so the algo­rithm basi­cal­ly flags where and when the next crime could hap­pen so the secu­ri­ty or the police offi­cers would deploy peo­ple to try to pre­vent that crime from happening. 

Or it could be people-based, where there are cer­tain pro­files devel­oped. So you devel­op cer­tain pro­files of like­ly attack­ers, and again the algo­rithm mon­i­tors peo­ple’s activ­i­ties online and if there’s some­thing that’s sus­pi­cious, it flags that per­son as a like­ly attack­er. And that sys­tem is what Israel uses. So they have built a num­ber of pro­files of like­ly Palestinian attack­ers based on cer­tain data points, some of it relat­ed to age, loca­tion, and the psy­cho­log­i­cal buildup of these peo­ple. And accord­ing to media reports there were psy­chol­o­gists sit­ting [in] on inter­ro­ga­tions to basi­cal­ly drag infor­ma­tion and see how do these young peo­ple think and what could be seen as some warn­ing signs.

But also they look for cer­tain trig­ger words which are often used in the Palestinian dis­course. Things like the word the ​“shahid,” which means mar­tyr, and this is a term that we use for any­body who dies from the con­flict; words like ​“Al-Quds,” which means Jerusalem; ​“al-Aqsa,” which is the holy Muslim site; any Quranic vers­es from the holy book; poet­ry; if you put a pic­ture of a per­son who just was killed or arrest­ed by the Israeli forces. And they par­tic­u­lar­ly look at the activ­i­ty of friends and fam­i­lies of peo­ple who just got arrest­ed, to assess whether these peo­ple would retaliate.

So for exam­ple, a 23 year-old Palestinian kid with Down syn­drome was shot dead and then two months lat­er they arrest­ed his broth­er. And dur­ing inter­ro­ga­tion, the inter­roga­tors showed him screen­shots of his Facebook and that he post­ed a pic­ture of his dead broth­er. And that was enough for them to arrest him and to ask him straight­for­ward, ​“Are you plan­ning a retal­i­a­tion attack against Israel, because you had the pic­ture of your broth­er up online?”

So what does this all mean, this pre­dic­tive polic­ing busi­ness for Palestinians? It means that if you’re a Palestinian liv­ing in Palestine, it means that you are a sus­pect by default and you could be arrest­ed for what­ev­er thing you post­ed online. And that this is for an imag­i­nary crime that you may or may not com­mit in the future.

And this is one of the things that some ex-veterans of… The elite unit of the Israeli army is called Unit 8200. It’s an elite sig­nal intel­li­gence unit that is part of the IDF and is often com­pared to the NSA in terms of its sur­veil­lance capa­bil­i­ties. So, some vet­er­ans or ex-veterans at the time, they wrote a protest let­ter to the Israeli Prime Minister Benjamin Netanyahu to protest against the uneth­i­cal, unlim­it­ed sur­veil­lance that is being com­mit­ted against Palestinians. And he says, and I quote, ​“There is no dis­tinc­tion between Palestinians who are, and are not, involved in vio­lence. Information that is col­lect­ed and stored harms inno­cent peo­ple.” So again this is to stress that pre­dic­tive polic­ing rein­forces this gov­ern­men­tal­i­ty that every­one is a sus­pect and every­one there­fore should be sur­veilled and monitored.

But for us here sit­ting in the room, we have to look at this pre­dic­tive polic­ing in Israel from a dif­fer­ent per­spec­tive, or a wider per­spec­tive. Because this unique com­bi­na­tion of hav­ing this algo­rithm as a polic­ing sys­tem, and also the abil­i­ty to detain peo­ple and put them under admin­is­tra­tive deten­tion, which I assume in Western democ­ra­cies that could not be the case—you can’t just simply…at least in the­o­ry, detain some­one and not press charges and not go through the prop­er legal process.

So that com­bi­na­tion is some­how unique to Israel and maybe can­not be repli­cat­ed else­where. But that of course does not mean that such tech­nolo­gies are not being employed and grow­ing in oth­er police and secu­ri­ty depart­ments around the world.

In fact, the EU Counter-Terrorism Coordinator (which I can nev­er pro­nounce his name) said in Israel in meet­ings with the Israeli gov­ern­ment and in some con­fer­ence that the EU is inter­est­ed in adopt­ing this pre­dic­tive polic­ing sys­tem used in Israel, specif­i­cal­ly to try and stop lone wolf for attacks in the EU. And in oth­er EU states like Denmark, the Danish police have bought a sim­i­lar pre­dic­tive polic­ing sys­tem from one of the com­pa­nies that alleged­ly is involved in the devel­op­ment of the sys­tem that we now know in Israel. And only recent­ly, like two months ago, the UK…there were some reports in The Independent that there is a del­e­ga­tion going from the UK gov­ern­ment to Israel to exact­ly coop­er­ate on pre­dic­tive polic­ing sys­tems after the recent attack on London Bridge. So again, they want to look at how to stop lone wolf attacks by mon­i­tor­ing social media and doing social media analysis.

So, of course that is not a sur­prise, giv­en that Israel is a leader in export­ing cyber arms, cyber­se­cu­ri­ty, sur­veil­lance tech­nolo­gies around the world. And I think the rea­son behind this promi­nence as the revolv­ing door between its army and the tech­nol­o­gy sec­tor. So, some of the lead­ing Israeli com­pa­nies that export sur­veil­lance tech­nolo­gies, or cyber arms, cyber­se­cu­ri­ty prod­ucts, they were found­ed by ex-veterans of Unit 8200. And there is no legal bar­ri­er for these peo­ple who while work­ing in this unit to take ideas devel­oped there relat­ed to mil­i­tary intel­li­gence and then lat­er sell them as pri­vate sec­tor com­pa­nies. Which means that, in the words of an Israeli schol­ar, the Palestinian ter­ri­to­ry’s become like a cer­tain lab to test cer­tain tech­nolo­gies and fine-tune them, and then sell them lat­er to coun­tries around the world.

And then of course the Israeli gov­ern­ment and oth­er gov­ern­ments jus­ti­fy the use of pre­dic­tive polic­ing by say­ing that it drops crime rates, or it’s effec­tive in stop­ping ter­ror­ism. But I think that these claims, just exact­ly like the sys­tems that they sup­port, they lack sta­tis­ti­cal evi­dence. So a lot of the research that was done on the effec­tive­ness of pre­dic­tive polic­ing shows no back­ing what­so­ev­er that they are indeed effec­tive. So one of the stud­ies, or actu­al­ly two, were done by the Rand Corporation in the US. They tried to look at the use of pre­dic­tive polic­ing sys­tems in cer­tain US com­mu­ni­ties, and that they found that they nei­ther reduced crime nor increased pub­lic safety. 

And there was anoth­er research done by the Human Rights Data Analysis Group to look at how pre­dic­tive polic­ing recy­cles exist­ing bias­es, some­thing that Maya just dis­cussed. They used algo­rithms and test­ed test in Oakland, US on drug crimes data, and they found out that the algo­rithm was telling them to go exclu­sive­ly to areas where there are only black res­i­dents and low-income residents.

So I guess, and I’m end­ing here now, that the argu­ment whether pre­dic­tive polic­ing is effec­tive or not is not what we need to think about. I think pre­dic­tive polic­ing changes the tra­di­tion­al log­ic behind law enforce­ment, which means that if you com­mit­ted a crime or you’re plan­ning to com­mit a crime and there is hard enough sub­stan­tial evi­dence to arrest you, then you’re pros­e­cut­ed. But I think we real­ly need to stop and and think very hard about sys­tems that change this log­ic and you might get arrest­ed and pros­e­cut­ed based on a crime that you may or may not com­mit in the future. Thank you very much.