Gisela Pérez de Acha is an open source investigative reporter specializing in extremism, AI, and environmental issues. She walked our 2026 participants through the key tools and methodologies to ensure human rights defenders are safe while both online and offline, as well as how to cultivate psychosocial resiliency. As OSINT trainer, she also introduced the importance of mastering open source investigation to strengthen research, reporting and accountability work.
Here the full interview.
What drew you from law into human rights work specifically; was there a case or moment that set the direction?
I used to be a human rights lawyer, and I became a journalist after the 2017 earthquake in Mexico City when I coordinated a group of almost 500 volunteers to verify “fake news.” Our goal: to save as many people as possible from underneath the rubble. After that I quit my job at an NGO and earned a Fullbright Scholarship to study a masters in journalism at UC Berkeley. So the question is not what drew me to human rights work, but rather, what drew me to journalism.
Looking back, is there a skill or perspective from your legal training that turned out to matter more than you expected in this field?
Once I got to UC Berkeley, I quickly realized that the Journalism School didn’t have a “digital verification” program. It was the Law School, under the brilliant Alexa Koenig and the Human Rights Center that she runs here I really learned the ins and outs of OSINT: open source investigations. Perhaps without the legal background it would’ve been harder to understand the framework that Alexa was coming from in order to translate it to journalism.
Cybersecurity and digital safety are now more than ever a priority for human rights defenders. Can you walk us through what you actually covered in your Venice School session?
We covered password managers and two-factor authentication first, since reused passwords are still the easiest way anyone gets into your accounts. Then separating your personal digital life from your work life, different browsers, ideally different devices if people can manage it, so a compromise in one doesn’t bleed into the other. We talked through metadata, what gets stripped automatically by apps like WhatsApp or Instagram and what doesn’t, because people assume everything is scrubbed and it isn’t. And we covered VPNs, not as some advanced spy tool but as a basic layer that also happens to be useful for your safety and not just your research.
None of it is glamorous. But for a lot of defenders, this is the gap. They’re doing serious, dangerous work and their actual attack surface is an old password and a phone with no lock screen.
Is there a skill you tell trainees NOT to outsource to AI, even as the tools get better?
Judgment. And writing, frankly. AI is very good for automating repetitive and tedious tasks, analyzing and cleaning data, creating apps, coding. It can help summarize a document, it can even help me estimate a shadow angle if I feed it the right academic papers first. What it cannot do is decide whether a source is credible, whether a claim is safe to publish, or whether pursuing a lead is going to get someone hurt. I tell my students constantly: do not work from memory, and I’d extend that to AI too. Never work from an AI summary alone.
On the other hand, AI-generated writing is just sad. I don’t know how else to put it. Writing is a thinking process, not a linear, mathematical summary of things. Writing should and must always stay human.
You are also an OSINT trainer: can you tell us more about this discipline and the importance, for the next generation of human rights defenders, to master these tools to strengthen their research, reporting and accountability?
OSINT is information that’s open to the public through observation, request, or purchase, nothing that requires special access or status. That’s the Berkeley Protocol definition and it’s the one I teach from. I think of the internet most people search as the tip of an iceberg, maybe one to four percent of what’s actually out there. OSINT is how you reach the rest: academic records, medical and legal databases, government filings, subscription content that a normal Google search never surfaces.
The discipline itself has a lineage. During the Cold War, intelligence agencies were already using open, publicly available material, radio broadcasts, newspapers, and treating it as real intelligence. Then the Arab Spring happened, smartphones spread, and a second wave of people repurposed those same techniques, not for espionage but to document human rights abuses in real time. That’s the tradition I situate myself in, and it’s the one I want the next generation to inherit.
For human rights defenders specifically, this matters because power tends to hide behind claims that evidence doesn’t exist or can’t be verified. OSINT gives you a way to build an evidentiary record independently, using nothing but public information and a rigorous method. It’s not about tools. It’s a mindset, closer to a detective’s than a hacker’s. The tools change constantly. The discipline of cross-referencing, documenting your chain of reasoning, and never trusting a single source doesn’t.
OSINT is over 7,000 tools, but you’ll very often hear me say that it’s mostly a method and a way of thinking.
Is there a case where open-source evidence changed the outcome of an accountability effort, one you can talk about?
So many. The most recent that comes to mind is one that my colleague Haley Willis shared during an OSINT Summit that we hosted together: a week-long online training session for journalists, security researchers and lawyers. She talked about how back when she worked at the New York Times, she documented what could very well be war crimes using OSINT in Bucha, a Ukranian town that was occupied by Russian soldiers in 2022. Satellite imagery and other OSINT methods were crucial to document a massacre around that time, and establish attribution to specific Russian battalions.
What’s the biggest misconception people have about what open-source investigation can actually achieve?
That it’s a shortcut, or that it’s spying. Neither is true. I actually try to avoid the word “intelligence” when I can, because it carries a surveillance connotation that doesn’t fit what journalists and human rights researchers are doing. We’re not trying to spy on people or weaponize personal information. We’re trying to verify what’s already public and build a case that can withstand scrutiny.
The other misconception is that more tools equal better results. I tell every cohort I teach: less tools, more methodology. Anyone can learn to run a reverse image search in an afternoon. What takes longer, and what actually matters, is learning to think like an investigator: to not trust your own memory, to put sources side by side, to notice when something’s too convenient. The tools will change every year. Some go dead, some get worse, some get replaced by something better. If your discipline depends on the tool instead of the method, you’re stuck the moment the tool disappears.
And what are the ethical challenges embedded in this work and methodology?
The line between verification and surveillance is thinner than people think, and it’s on the practitioner to hold it. Having access to something, a leaked password database, a private social media post, someone’s phone number cross-referenced across five platforms, doesn’t mean you have the right to use it. I tell my students plainly: we don’t hack, and having the technical ability to find something is not the same as having permission to use it, even against people you consider adversaries.
There’s also a real risk of causing harm through carelessness. Publishing an identification that’s wrong, or right but reckless, can get someone hurt, especially when you’re working on human rights abuses where the subjects of your research may retaliate. And there’s an emotional cost to the practitioners themselves. A lot of this work means looking closely at violent material, sometimes frame by frame, to verify what happened. I teach people to use tools like keyframe extraction specifically so they don’t have to watch an entire video of something horrific just to do their job. Protecting your own psychological safety is part of the ethics of this work too, not a side note.
If you could give one piece of advice to someone starting out in this field today, knowing what you know now, what would it be?
Learn the discipline before you fall in love with the tools. Pick one topic and actually practice on it, the way you’d learn a language by going to the country and forcing yourself to speak it rather than sitting in a classroom. And build the habit early of never working from memory. Write things down, double check with another source, put your findings side by side before you trust them. The tools you learn this year will be half-replaced in two years. The habits of mind will still be yours.