Le 11 août, Adam Bies s'est connecté à son compte sur Gab et a commencé à taper :

"Je crois sincèrement que si vous travaillez pour le FBI, alors vous méritez de mourir."

Bies, 46 ans, était un photographe indépendant en herbe qui avait rempli son site Web de photos d'action de voitures rapides et de sports de plein air. Il avait été licencié de son travail de jour dans le marketing pour avoir refusé le vaccin COVID-19, a-t-il écrit en ligne, et avait eu du mal à déposer une demande de chômage.

Comme les procureurs fédéraux le décriront plus tard dans les documents judiciaires, Bies remplissait ses journées en postant sous un pseudonyme sur Gab, un service de médias sociaux populaire auprès des extrémistes de droite.

Son message comprenait un lien vers un article de Fox News sur le directeur du FBI, Christopher Wray, dénonçant la vague de menaces violentes dirigées contre l'agence au cours des trois jours qui ont suivi la perquisition du domicile et du club de l'ancien président Donald Trump, Mar-a-Lago. Il a comparé les agents fédéraux aux forces nazies. Il a fulminé à propos de "l'écume de l'État policier". Et il a composé ce qui aurait pu être considéré comme un plan final.

"Je sais déjà que je vais mourir aux mains de ces … scumbags des forces de l'ordre", a-t-il écrit, entrecoupé de blasphèmes. "Mon seul objectif est d'en tuer plus avant de tomber."

Quatre jours plus tard, mandat en main, des agents fédéraux armés et des équipes SWAT ont encerclé la maison de Bies, près d'une chute d'eau tumultueuse dans le pays de chasse en forêt profonde de l'ouest de la Pennsylvanie. À l'intérieur de la maison se trouvaient Bies et son fils de 12 ans. Il faisait nuit, près de minuit.

Les agents ont appelé Bies sur son téléphone portable, encore et encore, 16 fois en tout. Ils ont donné l'ordre par haut-parleur de se rendre.

Enfin, Bies a émergé, portant un fusil d'assaut. Les agents lui ont ordonné de déposer l'arme.

Au cours de ces quatre jours entre les messages menaçants de Bies et le moment où il a affronté des agents armés, il avait été pris au piège par une pratique complexe et peu connue au sein du FBI appelée exploitation des médias sociaux, ou SOMEX, qui pourrait, à ce moment, surveiller les activités en ligne de n'importe qui en Amérique.

Les principaux dirigeants du FBI ont cherché à minimiser la mesure dans laquelle les agents peuvent légalement surveiller les activités publiques en ligne des personnes qui ne font pas l'objet d'une enquête. Mais en réalité, le bureau peut effectuer une surveillance presque illimitée des médias sociaux publics, tant qu'il le fait à des fins d'application de la loi, ont déclaré des responsables du FBI à U.S. TODAY.

Les experts disent que cela donne au FBI plus de pouvoir qu'il n'a voulu le reconnaître publiquement - pouvoir que le bureau et d'autres experts en sécurité disent qu'ils ont la responsabilité d'utiliser pour prévenir le terrorisme.

Mais les critiques affirment que l'exploitation des médias sociaux signifie également que les agents sont autorisés à examiner les publications en ligne à volonté, sans surveillance, mais avec de vastes pouvoirs.

"Les responsables du FBI ont diffusé de nombreuses informations erronées sur l'étendue de leurs pouvoirs", a déclaré Michael German, ancien agent spécial du FBI et membre du Brennan Center for Justice de l'Université de New York. "Le FBI a des pouvoirs énormes pour enquêter bien avant qu'il y ait un prédicat criminel raisonnable."

SOMEX, implique des agents qui développent leurs propres pistes et reçoivent des informations d'un réseau d'entrepreneurs et de collaborateurs, comme un groupe de recherche sur le terrorisme qui a d'abord signalé les messages de Bies.

Mais le bureau a été critiqué pour la façon dont ses enquêteurs ont réagi – comme dans le cas des messages en ligne publiés par des militants libéraux lors des manifestations de Black Lives Matter de 2020 – et pour la façon dont ils n'ont pas réagi – comme dans la montée de la droite vers l'insurrection du 6 janvier.

Le FBI fait depuis longtemps l'objet d'un examen minutieux pour excès dans la création de fichiers sur des personnalités publiques et autres, même s'ils ne faisaient pas l'objet d'une enquête criminelle. Et certains experts disent que l'agence a l'habitude de se concentrer sur des groupes de gauche comme les écologistes et les militants de la justice raciale, tout en ignorant les menaces des suprémacistes blancs et d'autres de droite. Ils disent que cette tendance se perpétue dans l'ère numérique.

Et les dossiers internes obtenus par un groupe de défense semblent montrer des agents de cyber-recherche se concentrant spécifiquement sur les rassemblements contre la police et la justice raciale au lieu de contre-manifestants armés ou de suprématistes blancs.

"Le problème avec la surveillance des médias sociaux est souvent le problème avec le maintien de l'ordre dans son ensemble, c'est-à-dire que la police ne peut pas prédire le crime, tout ce qu'elle peut faire est d'évaluer quel type de personne est le plus susceptible de commettre un crime et de mettre ce groupe sous surveillance. ", a déclaré Matthew Guariglia, analyste politique à l'Electronic Frontier Foundation. Cette "réaction instinctive", a déclaré Guariglia, finit par signifier plus de surveillance et de harcèlement des personnes de couleur et des groupes marginalisés.

Mais alors que l'indignation suscitée par Mar-a-Lago pousse désormais les menaces d'extrémistes de droite à des niveaux historiques, les questions de longue date sur la manière dont le FBI surveille réellement les Américains en ligne rencontrent une nouvelle tournure :que se passe-t-il lorsque les personnes menacées sont les agents du FBI eux-mêmes ?

Le FBI a plus de latitude que beaucoup ne le pensent

En juin de l'année dernière, lors d'une audition du comité de la Chambre sur la surveillance et la réforme, la membre du Congrès de New York, Alexandria Ocasio-Cortez, a interrogé Wray sur l'incapacité du FBI à prévoir le chaos de l'insurrection du 6 janvier.

"Nous savons maintenant que les attaques ont été planifiées au grand jour sur des plateformes de médias sociaux populaires", a déclaré Ocasio-Cortez. "Le FBI inclut-il régulièrement la surveillance des médias sociaux dans ses efforts de lutte contre l'extrémisme violent ?"

La réponse de Wray a été catégorique :

"Nous avons des politiques très spécifiques qui existent au département depuis longtemps et qui régissent notre capacité à utiliser les médias sociaux. Et lorsque nous avons un objectif autorisé et une prédiction appropriée, nous pouvons faire beaucoup de choses sur les médias sociaux", a déclaré Wray. . "Mais ce que nous ne pouvons pas faire sur les réseaux sociaux, c'est sans prédication appropriée et dans un but autorisé, il suffit de surveiller."

Des mois plus tôt, l'ancienne directrice adjointe exécutive du FBI pour la sécurité nationale, Jill Sanborn, avait donné une explication similaire au Comité sénatorial sur la sécurité intérieure et les affaires gouvernementales. "Nous ne pouvons pas collecter les activités protégées par le premier amendement sans passer par la prochaine étape, qui est l'intention", a-t-elle déclaré.

La sénatrice Kyrsten Sinema a poursuivi en demandant :"Donc, le FBI ne surveille pas les conversations sur les réseaux sociaux accessibles au public ?"

"Exact, madame. Cela ne relève pas de nos compétences", a répondu Sanborn.

Les propres règles du FBI disent le contraire.

Les responsables du FBI ont déclaré à U.S. TODAY que la déclaration de Wray était correcte, tout en reconnaissant qu'un "but autorisé" signifie simplement faire quoi que ce soit en conformité avec les devoirs d'un agent du FBI.

Cette "fin autorisée" est en fait extraordinairement large. La politique interdirait aux agents de consulter les médias sociaux pour, par exemple, garder un œil sur un partenaire romantique ou surveiller d'autres utilisations non policières. Mais cela permettrait à un agent de regarder essentiellement n'importe quoi en ligne, de manière proactive, si l'intention était d'arrêter un crime ou d'assurer la sécurité des Américains. Un responsable du FBI a déclaré que cela relevait de la "pénombre de la sécurité nationale, de l'application de la loi fédérale ou du renseignement étranger".

German, a fellow with the Brennan Center's Liberty and National Security Program, argued in a recent report that individual FBI agents have extraordinary leeway to look through public-facing social media posts without seeking authorization from their superiors in advance or even keeping an official record of their actions.

The FBI rules, laid out in their handbook and periodically updated Attorney General's guidelines, allow agents to conduct "pre-assessments" of possible threats, German said. Those pre-assessments can be conducted "without any factual basis to suspect wrongdoing," German writes in his report.

He and several other experts agree that the FBI certainly can, then, proactively monitor Americans' social media for signs of unrest, dissent or violence that might lead to criminal activity.

FBI officials told U.S. TODAY this is correct. There's no need for "proper predication," or evidence of a crime, when conducting a pre-assessment of a subject.

German's analysis of the rules was echoed by Brian Murphy, a former top FBI official who helped pioneer the FBI's social media exploitation efforts.

He cited Sanborn's statements, telling U.S. TODAY, "I just think that she was wrong." He said the agency has a risk-averse culture that prevents agents and managers from taking the steps necessary to fully protect Americans.

Sanborn, who is no longer at the FBI, did not respond to messages seeking comment. An FBI spokesperson said Sanborn's comments referred specifically to "conversations" on social media and not to public-facing posts by individuals.

While the bureau describes its authorities carefully, its agents—and third party contractors—can track critics of the government like Adam Bies, watching until their online rantings cross a line into outright threats.

Then the FBI can act.

What SOMEX really looks for

The FBI's SOMEX team, which sits within the agency's National Threat Operations Center in Clarksburg, West Virginia, receives and investigates tips on imminent social media threats from concerned citizens, other law enforcement agencies, independent monitoring organizations and others.

But the effort involves more than just acting as a catcher's mitt for incoming tips. It also develops its own social media intelligence.

Documents obtained by the open-government group Property of the People (and first reported by Rolling Stone) give insight into the broader social media monitoring role SOMEX plays inside the FBI. The documents detail reports from the team to federal and local law enforcement in the Seattle area during the civil unrest that unfolded in the wake of the murder of George Floyd.

"While overnight social media activity was very light, the SOMEX team did find some tweeting by individuals stating they would monitor police radio activity," reads a typical extract from the documents, taken from a June 2, 2020 situation report emailed to dozens of FBI agents.

"The FBI aggressively scours social media for information related to topics of Bureau interest," said Ryan Shapiro, executive director and co-founder of the nonprofit group, which provided U.S. TODAY with hundreds of pages of documents about the FBI's social media monitoring that it acquired through open records requests. "This routinely includes surveillance of Americans who are not the subject of an investigation or even suspected of committing a crime."

In a statement, the FBI said that SOMEX was created to assist in identifying "unknown subject, victim, or location information" when there's a threat to life by using publicly available information. The team then forwards information to the appropriate agency for further investigation and appropriate action.

FBI officials told U.S. TODAY that agents are not allowed to use specific SOMEX tools without additional training in privacy and civil liberties protections. Those tools include commercial software the FBI purchases to use in-house. The FBI also works with third-party contractors for social media analysis, the officials said.

One contractor is the private intelligence firm the Hetherington Group, which has trained law enforcement and the military on conducting online investigations.

Cynthia Hetherington, the firm's founder and president, said the company identifies "actionable intelligence" that can be used to protect life or someone's reputation by helping those it trains learn how to hyperfocus and efficiently identify a key collection of terms that demonstrate legitimate threats, such as "kill," "die," "shoot," "fire," "bomb," "rob."

"Individuals should be allowed to say what they want to say on the internet, but should also understand that it's open source and the parties concerned will trace it back" to them, Hetherington said.

Another way of saying that, said Shapiro, who holds a doctorate from the Massachusetts Institute of Technology focusing on government surveillance, is that the FBI can, and is, monitoring practically whoever it wants, whenever it wants.

"The FBI is almost entirely unhindered in its ability to monitor American social media postings," Shapiro said, "So when the FBI reported to Congress that it was unable to do so—I mean, that is a bald-faced lie. That's what the bureau does. They lie."

As the FBI becomes more interested in specific posts, the bureau can also ramp up its monitoring in more "intrusive" ways, FBI officials said. With additional internal approvals, FBI agents can access not just public-facing social media, but also private groups and chat rooms.

Even when accessing this more private information, the FBI's internal checks don't protect Americans' civil liberties, several experts told U.S. TODAY.

The FBI has a long and troubled history of focusing on groups on the left of the political spectrum while largely turning a blind eye to domestic extremists on the far-right, said Guariglia, who holds a doctorate in the history of police surveillance.

"Both historically speaking, and in current events, we've seen the amount of surveillance that has been marshaled specifically against groups fighting for racial justice increased exponentially than from what we've seen being monitored on the right," Guariglia said.

The FBI pushed back on this assessment. "The FBI aggressively investigates threats posed by domestic violent extremists," a bureau spokesperson wrote in a statement. "We do not investigate ideology and we do not investigate particular cases based on the political views of the individuals involved."

Are there enough resources to do the work?

The FBI isn't the only law enforcement agency doing social media exploitation.

The bureau's SOMEX team is part of a constellation of social media analysis that has occurred across the national security apparatus over the few years. The Department of Homeland Security has its own SOMEX team plus social media analysts at dozens of "fusion centers" across the U.S. sharing intelligence with local, state and federal law enforcement, said Mike Sena, executive director of one of those fusion centers, the Northern California Regional Intelligence Center.

The FBI also works to train and assist local police departments in their social media exploitation efforts, a tactic that came to light earlier this year in a report by the Intercept, which detailed how the bureau provided the Chicago Police Department with fake social media accounts to investigate demonstrators outraged at the Floyd murder by police officers in 2020.

The San Bernardino terrorist attack in 2015 turned out to be a "proof of concept" on the efficacy of social media analysis, Hetherington said, when reporting from Facebook to a fusion center social media analyst helped the FBI quickly identify the people involved.

But using social media analysis to identify future crimes, rather than research past ones, is a broader net. That federal effort to prevent crimes is still small given the scale of the internet, Sena said.

"Most people would be shocked in America," Sena said. "There's a small number of folks trying to deal with these threats that are huge."

Sena and Hetherington told U.S. TODAY that after the ACLU of California publicized law enforcement's use of commercial software to "monitor activists and protesters" in 2016, many companies stopped selling their software to law enforcement or minimized their capacity to use it to track online activity.

As a result, Sena said, "our people are manually doing things, they're doing the work, but they're having to work 10 times as hard as they used to."

That's why agencies plan to bring their teams together, at least virtually, to break up siloes and avoid duplication, Sena said. One byproduct of this effort, he said, will be fewer blindspots or gaps that can be used to accuse law enforcement of bias.

"Even if you're being proactive, it's basically walking with a teaspoon at a river and trying to put that in a bucket," Sena said. "We're not getting everything, but it's better than nothing."

But German argues in his report that the majority of social media exploitation work is actually counterproductive. The sheer volume of tips generated by contractors and the FBI's own analysts results in an "information overload," German writes.

"Obviously, the multiple forms of social media monitoring that the FBI and other law enforcement agencies conducted prior to January 6 was not helpful in preparing for the attack," the report states. "Yet after the Capitol insurrection, the FBI invested an additional $27 million into social media monitoring software, effectively doubling down on a failed methodology."

Ongoing investment in social media exploitation

Those efforts continue even in the weeks since the Mar-a-Lago search and backlash.

Three days after the FBI executed its Aug. 8 search warrant on Mar-a-Lago and was inundated by right-wing threats, Ricky Shiffer, a 42-year-old Navy veteran, walked into the FBI office in Cincinnati armed with a nail gun and an AR-15 rifle.

As U.S. TODAY reported, Shiffer had spent the last nine days of his life ranting on Truth Social, the social media company founded by Trump. His hundreds of posts included explicit threats against the federal government including "Kill F.B.I. on sight."

When his attack failed, Shiffer fled north along rural highways and into a standoff where was ultimately shot and killed.

The FBI said in a statement that it had been informed of Shiffer but that "the information did not contain a specific and credible threat."

Wray told the agency in a message the day after that attack that the FBI's security division would be adjusting its "security posture accordingly."

A $32,400 contract approved Monday—after discussion that started weeks before the search of Mar-a-Lago, Hetherington said—notes that the agency will hire the Hetherington Group to train its agents on SOMEX later this month.

According to a document the bureau filed to justify making the purchase without opening it up to bidding, "it is an immediate need to expand and broaden the social media knowledge for the NTOS SOMEX team." The FBI wrote that the training can provide it with expertise in the "forces and factors that lead to the radicalization of terrorism specifically white supremacy extremism."

That document was filed Aug. 11, the same day Shiffer carried a nail gun into an FBI office, then fled into the Ohio cornfields.

It was also the same day Adam Bies was logging post after post on Gab.

'Why don't you send them my threats'

As Bies tapped out his messages, he wasn't just speaking to his 1,600 followers. According to court documents, he also deliberately tagged Gab founder Andrew Torba in his posts, goading him to report Bies to the federal government.

"Why don't you send them my threats so that they'd at least have something credible to show on Fox News," Bies wrote in the post. "Just scrub my timeline for the posts you didn't delete after you threatened to ban me."

Also watching Bies' posts was a third-party media monitoring and analysis firm, the Middle East Media Research Institute. MEMRI cut its teeth monitoring Middle Eastern media for English-speaking audiences, but over the last three years has expanded to real-time social media monitoring, specifically for threats from white supremacists and other homegrown extremists.

"We're consistently in communication with (law enforcement and government) agencies at the local, state and national level, and providing" them with actionable intelligence, said Simon Purdue, director of MEMRI's Domestic Terror Threat Monitor team. "Having people like us helps speed things along."

MEMRI alerted the FBI, according to a later criminal complaint. The FBI contacted Gab, who handed over Bies' subscriber information and Internet Protocol logs for his computer connection. Soon, agents were outside his Mercer County home.

After a 30 or 40 minute stand-off at his home, Bies eventually emerged carrying an assault rifle, an FBI agent testified in court. Agents told him several times to drop the weapon, which he eventually did.

Had he not done so, the agent testified, according to local media reports, "It would have ended differently."

Bies' son left the house safely. Inside the home, agents found 12 other guns and a compound bow. Bies was taken into custody and charged under a law that covers making threats against a federal law enforcement officer.

He has pleaded not guilty and is being held awaiting trial. + Explorer plus loin

US plans for fake social media run afoul of Facebook rules

(c)2022 États-Unis aujourd'hui

Distributed by Tribune Content Agency, LLC