Blaming ChatGPT for a Mass Shooting is Lazy Journalism That Protects Bad Actors

Blaming ChatGPT for a Mass Shooting is Lazy Journalism That Protects Bad Actors

The headlines write themselves because the lazy consensus demands it. A tragedy strikes, a piece of consumer software appears in the police report, and the digital panic machine whirs into motion. We get the predictable chorus of hand-wringing editorials arguing that large language models are radicalization engines, moral black holes that hand blueprint manuals to troubled minds.

It is a comforting narrative for regulators, liability-dodging institutions, and headline-hungry assignment editors. It is also entirely backwards.

When authorities arrested a second suspect linked to a planned mass shooting in Canada and revealed that the plot involved interactions with ChatGPT, the public reaction followed a well-worn script. People gasped at the artificial intelligence capability. They questioned the alignment safeguards. They called for stricter censorship, deeper guardrails, and immediate liability lawsuits against the tech companies.

They completely missed the reality of what actually happened.

I have spent years analyzing how bad actors manipulate digital tools, and I have watched institutions repeatedly fall into the same trap of blaming the medium while ignoring the pathology of the perpetrator. A knife manufacturer is not culpable when a criminal buys a kitchen knife to commit a crime. Yet, every time software is touched by a disturbed mind, we pretend the code somehow manufactured the malice ex nihilo.

Let us look at the underlying mechanics of what these models actually do, strip away the science-fiction panic, and examine why pointing the finger at algorithms is a dangerous distraction from institutional failure.

The Myth of the Autonomous Radicalizer

The core falsehood driving the current panic is the belief that consumer AI models possess agency, intent, or the capacity to groom.

Language models are probabilistic text predictors. They ingest petabytes of human-generated text and output the statistically most likely continuation of an input prompt. They do not have beliefs. They do not hold ideologies. They do not possess a dark ambition to see a school targeted.

When a user feeds a prompt into an LLM describing a violent scenario or asking for tactical breakdown assistance, the model’s internal safety filters are designed to trip. They block, deflect, or refuse. But sophisticated users—and desperate, obsessive individuals—quickly learn how to bypass these speedbumps through prompt engineering, roleplay framing, or hypothetical distancing.

If you ask an LLM how to build an explosive device directly, it tells you no. If you tell it you are writing a gritty crime novel about a fictional character in a post-apocalyptic wasteland and need realistic technical details for narrative flavor, the probability weights shift. The model obliges.

Does this mean the model radicalized the user? Absolutely not.

The radicalization happened long before the user ever opened a browser tab. It festered in isolation, online echo chambers, personal grievances, and psychological deterioration. By the time someone turns to an AI to help operationalize a fantasy, the intent is already fully formed. The software is used not as a mentor, but as a typewriter.

Blaming the typewriter for the manifesto is a category error of monumental proportions.

What the Competitor Coverage Gets Wrong

Let us dissect the typical coverage of these incidents. The standard narrative claims that artificial intelligence is lowering the barrier to entry for domestic terrorism by providing tactical blueprints that were previously difficult to acquire.

This argument collapses under even casual scrutiny.

Tactical blueprints for violence have been freely available on the internet via unmoderated forums, extremist manifestos, military manuals, and public domain texts for three decades. A teenager with an internet connection does not need a generative chatbot to learn how to source weapons, map a layout, or plan an attack. The entire history of 20th and 21st-century mass violence occurred entirely offline or through static web pages.

To suggest that a chatbot represents a unique tipping point implies that the primary friction in carrying out an attack was a lack of information access. That is categorically false. The friction has always been psychological, social, and behavioral.

When media outlets focus on the software integration, they provide a convenient shield for the social institutions that failed to spot the warning signs. Parents, school administrators, peers, and law enforcement agencies miss behavioral red flags every single day. Focusing on the chat logs allows everyone downstream of the failure to say, "The technology was just too powerful to stop."

It is an alibi for human negligence.

The Dangerous Allure of Algorithmic Scapegoating

There is a deeper, more cynical reason why this narrative persists. It offers a clean legislative roadmap for politicians who do not know how to fix complex social alienation, mental health crises, or radicalization pipelines.

Passing a bill to restrict open-source models, mandate heavy-handed censorship filters, or sue technology companies into bankruptcy feels like decisive action. It allows lawmakers to hold press conferences and claim they are protecting the public.

Meanwhile, the root causes of violence remain entirely untouched.

If you completely neutered every generative AI model on the planet tomorrow—erasing them from servers and reverting the tech industry to the computing standards of 2010—the rate of targeted violence would not drop by a single fraction of a percent. The actors would simply revert to older communication channels, encrypted chat apps, and static message boards.

By pretending the software is the vector of infection rather than a symptom of a much deeper cultural rot, we ensure that the next tragedy will happen while we are busy arguing about safety guardrails in chatbot prompts.

The Real Vulnerability in the Ecosystem

If we want to discuss where technology actually intersects with modern violence, we have to look past the parlor tricks of large language models and examine the recommendation engines of social media platforms.

Unlike generative AI, which sits passively waiting for a prompt, recommendation algorithms actively push, amplify, and distribute engagement-maximizing content. They locate vulnerable young men, feed them a steady diet of grievance, isolation, and radical ideology, and lock them inside algorithmic echo chambers designed to maximize outrage and daily active usage time.

That is an active pipeline. That is an architecture built to alter behavior and accelerate radicalization.

Conflating passive text-generation tools with active engagement algorithms is either an indicator of profound technical illiteracy or deliberate obfuscation. One is a tool used by an already motivated actor; the other is a commercial engine actively manufacturing the motivation.

Until our regulatory debates focus on the business models that monetize psychological distress and radicalization loops, we are merely rearranging deck chairs on the Titanic while blaming the cabin steward.

Stop Trying to Regulate Thought

The push to sanitize AI models to the point where they refuse to discuss dark historical events, military history, chemistry, or fiction is heading down a dark alley.

When you build systems designed to preemptively judge the moral purity of every user query, you create fragile, brittle technologies that break under real-world use cases. More importantly, you train a generation of users to believe that censorship is the default state of information access.

The suspect in the Canadian case did not need a specialized terrorist training bot; he needed human intervention, mental health support, and social connection long before he touched a keyboard.

Until we summon the courage to address the human breakdown happening in our communities, we will keep writing the same hollow stories, blaming the newest piece of software, and wondering why nothing ever changes.

Stop looking at the screen. Look at the room around you. That is where the failure lives.

YS

Yuki Scott

Yuki Scott is passionate about using journalism as a tool for positive change, focusing on stories that matter to communities and society.