Exploits

We can essentially do mind control.

I work for a clandestine government organisation. I won't tell you which one, you know why. My background is in psychology and computer science and I run a laboratory where we experiment on human test subjects, mostly criminals. Our goal is simple: to uncover repeatable protocols that exploit the peculiar cognitive kinks nestled within the human brain, enabling us to influence, control, and even override rational thinking in subjects. After all, the brain, like any other complex software system, has its bugs.

Don't believe me? Think about it for a moment. We are not perfect, our perception and cognition are riddled with holes. Some we treat like toys, like optical illusions. Others we use for entertainment, like hypnosis. We are also open to manipulation, from being exploited to make purchases by sales and marketing messages, to conforming in schools and the military through structure and setting, and to joining religions and cults via confederates and systems optimised over millennia.

Do you, or do any of us, think the same or make the same decisions if we are hungry? If we are angry? If we are in love? Emotional affect is an easy lever to pull.

We can bring machines into play, use neuromodulation and trigger almost any part of the brain, brute force specific emotions in the target, like feelings of bliss or terror. But we try to avoid interventions that require gadgets. We prefer good old-fashioned brain hacks.

We develop protocols that agents can use in the field to predictably make use of these defects. Once we find an attack vector, we then test different ways to exploit it, trying to find the minimal number of inputs to achieve the maximum effect.

Let me give you some examples from our lab.

This first one is pretty tame. We focus on the phenomenon of visual afterimages using a grid of lights, expertly programmed to flash intermittently, with brief intervals of complete darkness in between. As we expose our subjects to this mesmerising sequence, a peculiar phenomenon unfolds. Instead of visions dancing before their eyes, a vivid image of an ethereal entity, a spectre, is etched onto their retinas. This spectral apparition clings to their vision for minutes afterwards, lingering like a ghostly imprint. It is quite unsettling to watch, as our subjects are negatively affected, to say it mildly, their psyche subtly altered by the encounter. This is a hard one to use in the field though.

It's not just getting people to see things, we can get them to do things, like precipitate actions on command. Actions like picking up a gun and pulling the trigger. For example, we developed a procedure using little more than body language, priming, and carefully designed subliminal cues. With these tools, we can manipulate the subject and bypass all of their rational safeguards. Of course, we used rubber guns while optimising the procedure. It took many months to work out the kinks. This one has been employed in the field. In fact, there was a case only last month in the mainstream news that you might have seen.

However, techniques advance. We have started to introduce the power of artificial intelligence into our experiments. Initially, we used AI to assist in the optimisation process of new procedures to improve efficiency, which proved very effective. We dropped from months to days to bring a new vector from proof of concept to something we could use in the field. The trick was the use of precise challenge experiments proposed by the AI.

These modern language models can tirelessly ingest and crunch data, assimilating vast volumes of scientific papers, and in turn, provides us with a wealth of innovative insights. The AI has the ability to make connections across broad fields of science. Surpassing our wildest expectations, it can generate a multitude of novel attack vectors that we would have never conceived on our own.

Let me share a few details of one of our current endeavours involving artificial intelligence. I have to hold back on specificity due to the sensitive nature of our work, but it highlights the remarkable potential that AI holds. During our exploration, the AI stumbled upon a clever hack within the processing of language in the brain. With the short repetition of certain language-like sounds, we found ourselves able to exploit this newfound vulnerability to a startling degree. This is not something tailored to the subject, it seems general.

We can achieve remarkable results, like having the target speak complete sentences, yet be left with no recollection of their words, replaced instead by an overwhelming sense of contentment. By repeatedly activating specific regions within the auditory cortex, we overwhelmed the semantic processing part of the brain, effectively bypassing critical thinking and granting us a fleeting moment of control. In those few precious seconds, the subject becomes a puppet, obediently enacting every command, like "stand up and jump" or "say the following". It's more effective than hypnosis, it bypasses the subject's will and memory directly.

The profound impact of this discovery cannot be understated. We have stumbled upon a tool of immense potency, one that had eluded our grasp until AI came along. It's simply marvellous. Of course, it's early days. We have not tried this procedure in the field yet. We think this is just the tip of the iceberg.