From Blogging to Belling the Cat: How Eliot Higgins and Bellingcat Became the Vanguard of Digital Sleuthing in the AI Era

To review this article again, go to My Profile and then select Saved stories.

By Samanth Subramanian

Guiding a Brigade of Internet Detectives in the AI Era

A decade prior, Eliot Higgins could enjoy in-room dining at hotels without the slightest worry about being targeted with poison. At that time, he hadn't been labeled as a foreign agent by Russia; actually, he wasn't even on the radar of security agencies in Russia or anywhere else. He was simply a British individual stuck in a mundane office job, who had been writing on his blog under the pseudonym Brown Moses—a reference to a song by Frank Zappa—and was in the midst of transforming his blog into a fully operational website. He was essentially a pioneer in the field of open source intelligence, meticulously examining social media images and videos along with other online debris to uncover war crimes in Libya and Syria.

Through its chaotic nature, the internet provided him with ample proof, allowing him to reach conclusions even before UN investigators could. This led him to take his efforts to a professional level. He named his website Bellingcat, inspired by the story of the mice who devised a method to detect the imminent threat of their predator. He aimed to be the mouse that alerted others to danger.

Currently, Bellingcat stands as the leading agency globally specializing in open source intelligence. Operating out of his UK residence, Higgins manages a team of almost 40 individuals. This team employs a continuously advancing array of digital forensic methods to delve into incidents ranging from the 2014 downing of Malaysia Airlines Flight 17 over Ukraine, a dog theft in 2020, to multiple assassination attempts on Russian critic Alexei Navalny.

Bellingcat functions as a non-profit organization based in the Netherlands but its services are sought after globally. The team provides training to news organizations and runs workshops; they expose war atrocities; and their analytical evidence is becoming a regular feature in legal proceedings. During a meeting with Higgins on a Saturday in April, at a tavern close to his residence, he shared that he had just returned from the Netherlands after receiving a prize recognizing Bellingcat's efforts towards free expression—and was preparing to return to accept yet another accolade, this time for their work in peace and human rights.

The story of Bellingcat reveals a harsh reality about the quest for truth in today's world. When Higgins started his journey as a blogger under the alias Brown Moses, he was well aware of the internet's dark sides. However, he, like many journalists globally, has come to realize that the public's judgment is flawed. The importance of solid evidence has diminished; in the digital realm, everyone has the freedom to craft and believe in their own versions of reality, even if they are built on falsehoods. In addition to unearthing the truth, Higgins has been on a quest to find environments where truth is valued and has the power it’s meant to have — to uplift the oppressed and to bring the wrongdoers to justice.

The upcoming year could potentially be the most significant one yet for Bellingcat. The organization is not only monitoring conflicts in regions such as Ukraine and Gaza but is also dealing with an influx of fake materials related to elections in the United States, the United Kingdom, India, and many other nations. On top of that, the rise of artificial intelligence presents a new challenge. Although AI technology is currently not advanced enough to deceive the experts at Bellingcat, it's becoming increasingly capable of misleading the general public. Higgins is concerned that governments, social media companies, and technology firms are not taking this threat seriously enough. He fears that they will only recognize the true dangers of AI-generated content after it leads to a significant incident that causes tangible harm, suggesting that by then, it might be too late to act.

Authored by Matt

Authored by Matt

Authored by Megan Farokhmanesh

Authored by Joseph

WIRED: You're at the helm of the biggest open-source intelligence agency run by citizens across the globe. When you transitioned from blogging to founding Bellingcat's website ten years ago, what direction did you envision it heading in?

ELIOT HIGGINS: At that stage, I was still in the process of figuring out how to transform this into a viable career. I had been running a blog for a few years already. However, with children to support, the need to make a sustainable income became more pressing. The inception of Bellingcat was driven by the desire to create a platform where individuals could share their own content. By then, I had received several requests from people wanting to post on my blog. A more professional-looking website was necessary. Moreover, I envisioned it as a gathering place for contributors. However, that was as far as my planning went. There wasn’t any elaborate strategy in place. It was more about navigating what was immediately ahead, like planning for the coming week.

I initiated my project on July 14, and then, just three days afterward, MH17 was tragically brought down. The incident significantly propelled the open source investigation movement, both in the rapid expansion of our community and in the refinement of our investigative methods. It also greatly elevated our visibility. Currently, our Discord server boasts over 28,000 members. This platform enables individuals to share potential investigative leads, and our published articles often stem from these community contributions.

Life has been anything but dull lately. How have things been at Bellingcat since October 7, for instance?

We have expanded our team by employing additional staff. Our strategy includes integrating more editors into our operations. We have also reallocated employees from various initiatives to enhance our capacity. Currently, we have designated an individual solely responsible for managing video archives. However, the nature of the content we're receiving, unlike what we've seen from places like Ukraine or Syria, is considerably reduced in terms of on-the-ground footage.

Due to disruptions in internet connectivity?

Indeed, much of the evidence we come across involves misconduct by Israeli soldiers, actions that can certainly be considered breaches of international regulations. Interestingly, this information is often disclosed through their own social media profiles, where they share it voluntarily.

A further problem arises due to the electricity shortage in the area, leading to many events occurring under the cover of darkness, which aren't visible in recordings. For instance, the drone-captured convoy assault by Israel – there's plenty of video evidence, but it's engulfed in darkness. However, I came across an insightful analysis where experts utilized the sound from the footage to identify the types of weapons deployed. The audio signatures of the weapons were quite distinctive, enabling this identification.

Sounds similar to the unique acoustic imprints of

Indeed, recognizing the weapon's model is just one aspect; listening closely, one can discern not only the whizz of the bullet but also the distinct sound produced by the gun's barrel. This auditory cue can help determine the distance of the shot. In the tragic incident involving Shireen Abu Akleh, an Al Jazeera reporter who lost her life in 2022, available footage revealed she was fired upon from an area under Israeli control. Subsequently, months following the incident, the Israel Defense Forces acknowledged a "high possibility" that the fatal shot had been fired by one of their own soldiers.

Authored by Matt

Authored by Matt

Authored by Megan Farokhmanesh

Authored by Joseph

Are there new developments emerging from this conflict that you've yet to witness?

For the first time, I've observed artificial intelligence-generated material serving as a justification to disregard authentic information. The common perception about AI is its potential to deceive individuals into accepting falsehoods. However, the reality is somewhat different; it's essentially enabling individuals to doubt the authenticity of genuine information. This is because they can easily dismiss it by claiming, “That’s just something created by AI. Nowadays, AI has the capability to produce anything, be it images, sounds, or even simulating an entire battle scene.” It has become a convenient scapegoat for skepticism.

Subsequently, they are able to remain within their personal bubble of information…

Certainly, while browsing through your social media timeline, you have the option to quickly ignore content that doesn't align with your views, which only serves to solidify your existing opinions. Given the prolonged and complex nature of the Israel-Palestine conflict, there's already a significant number of people ready to be emotionally involved. Consequently, you witness opportunists producing content with falsely attributed images or ones created by artificial intelligence. This results in a very poor level of discussion. Consequently, finding genuine responsibility in this realm becomes challenging.

In this fully open approach, all the evidence and investigative work is uploaded online for public scrutiny, allowing anyone to verify it. However, it's a characteristic of today's society that individuals with fixed beliefs tend to cling to those beliefs despite overwhelming evidence to the contrary. Does the challenge of altering opinions bother you?

Regrettably, it's become something I've adapted to. That's the reason we're shifting our focus towards ensuring legal responsibility and exploring ways to leverage open-source evidence to achieve that goal. Our dedicated team is specifically concentrating on this aspect. Possessing the truth is one thing, but without holding individuals accountable, the truth loses its significance.

How do you define being held responsible under the law?

Individuals on site are documenting instances of war crimes. The question arises: how do we transition this evidence from being viewed on YouTube to being presented in a legal setting? There hasn't been a precedent where someone has brought a compilation of open-source evidence before a court demanding its consideration. Consequently, we’ve initiated simulated court cases, utilizing information gathered from probes into Saudi-led bombings in Yemen.

Much of our efforts involve teaching: Generally, attorneys are not well-versed in the realm of open source investigation. It's essential for them to be educated on the methodologies of investigators, their objectives, and how to discern flawed analyses.

The quality of analysis using publicly available data is deteriorating. Have you heard of Nexta TV? This media outlet from Belarus posted a series of tweets following an incident at a concert in Moscow. They pointed out the presence of numerous individuals in blue sweaters, suggesting they might be operatives of the FSB [Russia's Federal Security Service]. However, they failed to provide any concrete evidence to support their claim that these individuals were indeed FSB agents. This poorly conducted analysis gained traction and misled many into believing there was a covert operation underway. It seems that as long as someone can highlight something with colorful boxes and claim it's part of an open-source investigation, there will always be those ready to accept it as truth.

This year, the United States, the United Kingdom, and India are all scheduled to hold elections. Are you gearing up to manage these three major electoral occurrences alongside the situations in Ukraine and Gaza?

Authored by Matt

Authored by Matt

Authored by Megan Farokhmanesh

Written by Joseph Cox

Our capacity to get ready is limited, as the onslaught of misinformation and artificial intelligence-created visuals is expected to be overwhelming. Observing the current events in the US primaries illustrates this point, with instances like fraudulent automated phone messages already surfacing; notably, the DeSantis team utilized artificial intelligence to create images depicting Trump and Dr. Fauci embracing. This indicates that the boundary has indeed been breached. Moreover, these technologies are not exclusively in the hands of political operatives but are accessible to the general public too.

This greatly exacerbates the situation.

Indeed, it's not solely about the actions of the campaigns themselves, but rather the choices made by their backers.

With the surge of artificial intelligence-produced images, do you have concerns about Bellingcat simply becoming a verifier of facts instead of conducting more in-depth probes where you compile evidence?

The recent situation involving Kate Middleton serves as a prime example. Initially, I resisted the urge to engage in what I considered to be a pointless discussion. However, as I noticed numerous TikTok videos claiming that the color in images was altered, each garnering millions of views, I felt compelled to contribute my thoughts. This incident perfectly illustrates the genesis and proliferation of false information and its underlying mechanisms.

I must admit, my interest was also piqued, albeit briefly, for about a week.

This is precisely why it became a hotbed for misleading information! In my experience with numerous groups inclined towards conspiracy theories, it's clear that they don't see themselves as proponents of conspiracy theories. Instead, they view themselves as crusaders for the truth, standing up to some form of power that they feel is deceiving everyone. Based on their own experiences, they've reached the conviction that established authorities are not reliable.

I recall a term you once coined beautifully: individuals who subscribe to conspiracy theories have experienced a form of "traumatic ethical harm."

Authored by Matt

Authored by Matt

Authored by Megan Farokhmanesh

Authored by Joseph

I often refer to Covid as an example. Many individuals spreading misinformation about Covid came from the alternative health circles, having previously encountered negative interactions with healthcare providers. This includes instances where treatments failed, loss of a loved one occurred, or they faced poor treatment. Indeed, some of these experiences are genuine and involve real trauma.

Recently, they discovered individuals who shared their perspectives, forming a community that includes members opposed to vaccinations. With the arrival of Covid, these dissenting opinions gained significant volume within these groups. The existing skepticism towards healthcare experts became even more pronounced. This situation is about amplifying their fears, which are constantly fueled each time they navigate through their social circles.

In a time where images created by artificial intelligence are set to become widespread, isn't it preferable for individuals to develop an enhanced sense of caution towards their surroundings, to be more doubtful than excessively credulous?

I would challenge the premise of that question. Constantly triggering people's intuition of danger leads to pervasive skepticism. This phenomenon has been observed in the context of Israel and Gaza. Many individuals are deeply concerned about the unfolding events, yet the complexity has reached a level where they feel compelled to disengage. This alienation of the moderates in the discussion poses a significant risk to democratic societies that rely on the ability to engage in open debate, doesn't it?

Has AI-generated content reached a level of refinement that makes it challenging even for your group to differentiate it?

In our investigation, we delve into the intricate web of details linked to a photograph. During the validation phase, we examine a multitude of data points. The initial step involves confirming the photograph's geographical location. We also scrutinize elements like shadows, which can help determine the time it was captured; essentially, knowing the camera's position allows us to use it as a makeshift sundial. Furthermore, we assess the photo's embedded metadata. Once the image circulates online, we investigate the uploader's social media profile, paying close attention to their connections. This could reveal acquaintances in the vicinity who might have witnessed the same event.

AI-generated visuals have the power to influence significant events, exemplified by the AI-created image of the Pentagon that momentarily impacted the stock market. [A photo emerged in May 2023 depicting a massive smoke cloud over the grounds of the US Department of Defense.] Under normal circumstances, such a major event would be reported by various news outlets almost immediately. It's unlikely that something of this magnitude would go unnoticed. However, this time, there was just a single source for the image, indicating it was a fabrication.

I'm worried that at some point, someone will put together a plan involving coordinated attacks through social media, utilizing networks of bots and disinformation sites that have established credibility over time to craft a storyline. If someone smart decides to produce a vast array of counterfeit content and distribute it simultaneously through these channels, claiming an event has occurred, it could cause significant confusion. This confusion might be enough to influence the stock market, trigger widespread panic, or even lead reputable news outlets to unintentionally report on it, exacerbating the issue further.

Authored by Matt

Authored by Matt

Authored by Megan Farokhmanesh

Authored by Joseph

Where do we start in addressing this issue?

It is imperative that social media platforms are mandated by law to incorporate artificial intelligence mechanisms for identifying and signaling content during the publication phase. Merely having a fact-checking feature is insufficient. Relying on voluntary compliance from these companies is unlikely to be effective. There must be repercussions for their failure to implement such measures. My concern is that the urgency of this issue will only be recognized following a significant adverse event.

"Many individuals experience a phenomenon known as compulsive witnessing. They feel an overwhelming need to observe something because, historically, there have been instances where people have chosen to ignore significant events."

Are you currently engaged in a significant amount of investigative tasks on your own?

Certainly. Whenever there's a free moment in my schedule to tackle a brief geolocation task or similar, I seize the opportunity. My engagement spans extensively across the operational aspects of our production business, which occupies a lot of my time. Additionally, I contribute significantly to public relations and communications efforts.

Does that come naturally to you? I recall you mentioning once that in your younger years, you experienced a bit of social anxiety, is that correct?

I struggled profoundly with social anxiety. It was a battle to overcome it. Initially, my anxiety was overwhelming, especially at the thought of public speaking. The notion of standing on stage filled me with dread. My first major public speaking event was at a Google Ideas summit in 2013, an experience I recall only as being overwhelmed with nerves. However, repeatedly facing this fear, particularly for causes close to my heart, has significantly mitigated it.

What activities occupy your leisure moments on the internet? How do you like to pass the time during vacations?

I've deleted Twitter from my mobile device, since it was among the most negative aspects. Engaging in disputes with individuals…

I've observed that you've stopped doing that. It was something you did frequently and with genuine intent before.

It felt similar to evaluating my own understanding. If an individual approaches me to point out a mistake in my reasoning and I find myself unable to counter their argument, I need to accept that I'm mistaken. Engaging in these discussions used to be beneficial, regardless of whether the opposition had sincere intentions or not. However, the situation changed when the myths surrounding Bellingcat solidified within certain circles. Nowadays, when someone comments, “Bellingcat is just the CIA,” it's always the same baseless claim.

Alright, if the arguing has decreased, what other activities are you engaging in?

I frequently employ artificial intelligence for personal amusement. Are you familiar with Suno AI or Udio? They're platforms designed for making music, and in the last half-year, they've made significant advancements.

Ah, Suno. This term in Hindi means "listen."

Have you had any experience using these?

Certainly not.

Let me demonstrate. I maintain a SoundCloud profile for sharing my musical creations. You have the option to suggest musical styles. Additionally, you can submit your own lyrics.

Here's what the creator of Bellingcat does in his free time.

Certainly. I'm particularly fond of the moments when the AI generator goes on a tangent, veering into the bizarre. I've penned numerous songs centered around concepts like online echo chambers and similar topics. I've discovered that distilling an idea into a verse format aids in breaking it down into a more digestible form, making it easier to convey to others through articles and books.

Are the prompts you're providing influenced by specific styles, or are they merely categorized by genre?

I've developed a whole method for this now! Initially, I would just think, "Let's create a song with an ambient vibe." However, I started to ponder: How can I achieve the distinctive sound of specific bands? Because simply instructing it to "Produce a song like Beastie Boys" doesn't work; the system doesn't accept prompts in that format, likely as a precaution against legal issues. So, I turn to ChatGPT and describe my situation: I need to come up with style tags and music genres for a music-creation software, asking for instance, what would be the style tags for a band like Kraftwerk? ChatGPT then deconstructs the music styles into individual tags, which I can then use and input back into the program.

Authored by Matt

Authored by Matt

By [Your Name]

Authored by Joseph

I've come across mentions that you refer to the longing for the pre-internet era as "cyber-miserabilism," which is a fantastic term. However, it's also accurate to say we all recall feeling more at peace mentally before we began browsing through various feeds.

Nowadays, you're always connected. My concern is the psychological harm this is causing individuals. This was particularly prevalent during the Ukraine crisis in 2022, where a vast number of individuals were constantly consuming news updates. Many expressed feeling perpetually distressed. It became apparent that we were unknowingly inflicting psychological damage on ourselves. A similar problem is occurring with the Israel and Gaza conflict, where the endless flow of images serves only to deepen the animosity felt towards the opposing side.

During Bellingcat's initial period, you encountered such videos regularly, frequently showing images of deceased individuals. How do you safeguard yourself against the impact of these visuals?

To me, it seemed meaningful since I achieved progress by witnessing everything. It's the feeling of helplessness that frequently accompanies a traumatic reaction. However, you can teach yourself to detach from it.

Is it possible, however?

I believe I've become quite adept at separating different aspects of my life. This skill is crucial for my profession. In the case of MH17, while examining the crash site's remains, I encountered a large, detailed image. As I inspected the shrapnel punctures, I noticed a doll amidst the debris, identical to one my daughter received from her aunt at birth. At that moment, an involuntary emotional connection forms. It's essential to pause there, as attempting to ignore it can have detrimental effects.

While examining those affected by the sarin gas attacks in Syria in 2013, our aim was to pinpoint the signs of exposure. Notably, a key sign is pupil constriction. This required me to scrutinize the eyes of deceased individuals closely to gather sufficient visual evidence to determine how they died. This task was distressing on its own. However, the situation worsened when encountering individuals online who dismissively claim the incident was fabricated, denying any real fatalities and absurdly suggesting that infants involved were pretending. Such assertions add an additional layer of trauma.

Many individuals experience an overwhelming urge to be observers, driven by the thought, "I must see this event unfold." This stems from historical instances where individuals chose to ignore what was happening. Thus, they feel a duty to acknowledge the hardship of others, believing it somehow restores a sense of control over the circumstances. However, this belief is misleading since it doesn't actually alter the situation. The only outcome is that it ends up causing them psychological distress.

It's been mentioned that Bellingcat provides mental health assistance, allowing team members access to complimentary counseling services. Is this therapeutic support widely utilized by the staff?

Absolutely, without a doubt. It's not only the content we encounter that's the issue, but also the response from governmental bodies, which, as you're aware, can be pretty forceful.

By [Your Name

Authored by Matt

By Megan Farokhmanesh

Authored by Joseph

I was curious about something. I've learned that you've stopped ordering meals from room service, and I'm interested in finding out what other habits you've adopted or abandoned. Additionally, how did things change for Bellingcat after it was labeled a foreign agent by Russia in 2021?

Our security team regularly engages in extensive discussions and reviews focused on both cybersecurity and physical security measures. We also organize staff retreats, during which experts provide us with guidance on personal safety protocols, such as actions to take if one finds themselves being followed, among other intriguing topics. Being labeled as undesirable or a foreign agent carries a dual significance for us. On one hand, it's somewhat of an honor; on the other, it presents challenges. Our commitment to transparency includes openly sharing information about our funding sources. However, this designation complicates matters, as acknowledging contributions from individuals associated with Russia could endanger them. Consequently, we have reluctantly ceased to disclose some of our donors' identities to safeguard their privacy and security.

Regarding this particular gathering, for instance. How were you able to decide to accept my invitation for a coffee? What actions did you take?

To begin with, I conducted some background checks. It was important for me to recognize your appearance since there have been cases where individuals find themselves in discussions with reporters, only to be caught off guard by unexpected and peculiar inquiries. These journalists might abruptly express negative opinions, such as criticizing Israel, leaving you puzzled about their intentions. I'm aware of instances where individuals were engaged in Skype conversations that were later manipulated and broadcasted on Iranian state media without their consent.

I came across a statement on the internet made by one of your ex-employees, where he mentioned, "Data serves as a powerful balance between a person and the government." However, isn't it inevitable that eventually, state bodies and spy organizations will improve at concealing their information?

Russia attempted the same strategy. Following our initial probe into the poisoning [of Sergei Skripal, an ex-Russian spy, in the UK], we obtained their GRU paperwork. However, when we made a second attempt, we noticed they had eliminated pictures from the GRU officials' documents. This, inadvertently, confirmed their association with the GRU. After we highlighted this fact, the images reappeared, albeit with faces of entirely different individuals. In a notable mistake, they switched a male's photo with a female's. This clearly demonstrated a lack of intelligence on their part.

However, they're inevitably going to become more intelligent?

Potentially. The reality is that these are simply entrances. If one entrance is blocked, we have the option to explore any of the other 10,000 available paths. The inquiry is far from over. We must simply choose a different path forward.

Share your thoughts on this piece with us. Send your feedback via email to the editor at mail@wired.com.

Recommended for You …

Direct to your email: Subscribe to Plaintext—Steven Levy offers an in-depth perspective on technology.

Within the largest undercover operation ever conducted by the FBI

The WIRED Artificial Intelligence Elections Initiative: Monitoring over 60 worldwide electoral events

Ecuador finds itself utterly without electricity due to a severe drought.

Be confident: Here are the top mattresses available for purchase on the internet

The individual is named Nilesh

Hannah Zeavin

Brendan I. Koerner

Steven Brill

Thomas Dexter

N/A

Additional Coverage from WIRED

Evaluations and Manuals

© 2024 Condé Nast. All rights reserved. Purchases made through our website may result in WIRED receiving a commission, as part of our affiliate agreements with retail partners. Reproduction, distribution, transmission, storage, or use of the content from this website is strictly prohibited without explicit prior consent from Condé Nast. Ad Choices

Choose a global website