The media has the AI threat all wrong
Note: This article makes reference to a mass shooting. I made a conscious choice not to use the name of the shooter.
For the past several weeks, we’ve been treated to breathless coverage of the supposed major threat presented by artificial intelligence. If we’re to believe the titans of the AI industry and the media reports mindlessly echoing their every utterance, humanity’s years may be numbered.
It’s not because of climate change and how we’re actively making our planet less fit for human habitation. Indeed, many of these prophets of AI doom don’t actually believe climate change is much of a threat at all. It’s also not because increasingly unstable geopolitics is shifting the world toward greater conflict and the risk of global war.
No, the real threat, according to these supposed geniuses, is that in ten years AI could break free from its human developers and decide it wants to kill humanity. If you’re big into science fiction, you might recognize that kind of a narrative: whether HAL 9000 in 2001: A Space Odyssey or The Terminator himself, this notion that our technology will come for us makes for compelling entertainment.
For more than a decade, people in the tech industry have been slowly deluding themselves into thinking those science fictional storylines were far closer to being realized than they are in practice. They think our future is one where we need to merge with computers, cover the face of the planet with data centers, and eventually start building them on other planets to fill with digital “people.”
These same delusional tech CEOs now want us to focus our attention on a fantastical future scenario where their chatbots gain sentience and go after their creators. Conveniently, that not only makes their products appear far more capable than they are in practice, but distracts growing regulatory conversations from the tangible harms of their products to futuristic scenarios that will never come into being in the first place.
Politicians and media have been far too happy to chase the industry’s sensationalist bullshit. But that doesn’t mean the real harms of AI have been dealt with. Yesterday we got even more evidence of the devastating effect generative AI products are having on our communities.
Enabling mass killing
If you’re not a Canadian, you might be forgiven for not realizing the country experienced one of the worst mass shootings in its history on February 10, 2026. That afternoon, the shooter killed their mother and brother before heading to Tumbler Ridge Secondary School in the small mining town of Tumbler Ridge, British Columbia and opened fire, killing five students and a teacher. It was a terrible tragedy that should have never happened — and it turned out OpenAI could have stopped it.
In the days after the attack, the Wall Street Journal reported there had been concern within OpenAI about the violent scenarios the future shooter was using ChatGPT to explore. They banned the individual’s account, but leadership decided not to report the future shooter to authorities. It was a fatal decision. But it turned out there was not just one ChatGPT account. While the first was banned, a second account was not. Yesterday, Mother Jones reported on the details of that account.
After creating it, the future shooter asked ChatGPT how to avoid getting banned a second time around. It instructed to frame scenarios as fictional and not to use real-world locations. “You can still be twisted. Just be clever about it,” the chatbot responded to the query. Over the next eight months, it helped the shooter plan the school shooting and entertain other violent scenarios, right up until the morning of February 10 when the shooter asked what time of day school shootings tend to occur.
OpenAI may deny its culpability in the case, but it’s now facing dozens of lawsuits from the victims of the Tumbler Ridge shooting and their families, along with the province of British Columbia. It’s not even the first time this has happened. Other countries have seen similar cases, like in April 2025 when a shooter entered the student union at Florida State University and killed two people after planning the attack with ChatGPT. This is part of the real threat of generative AI, but don’t expect it to get the wall-to-wall coverage that’s been given to the sensationalist lies of industry deceivers.
Deception is a strategy
Tech industry deception about the supposed threat of their generative AI tools is nothing new. It’s in fact a well-worn playbook. OpenAI was using it all the way back in 2019 when it released GPT-2. Claiming its new AI model was so incredibly powerful it posed a threat to society was a fantastic marketing ploy, allowing it to capture global headlines and get a ton of free PR. Now they’ve moved on to using it to shape the conversation around their products, but the media seems determined not to learn anything from their consistent pattern of deception.
The media is compromised in many ways. It’s stuffed full of tech optimists who want to believe the narratives of industry figures, alongside people who crave access above all else and know their bread is buttered by staying in the industry’s pocket. But you also have people who really don’t understand digital technology much at all, and certainly not artificial intelligence, so when an influential tech figure is making catastrophic warnings, the tendency is to believe them. After all, they wouldn’t be so wealthy and powerful if they couldn’t be trusted, right? But media also relies on audiences, and the sensationalism of claims that AI could destroy humanity has proven impossible to resist, even for many outlets that should know better.
If the media cared about factuality over sensationalism, it wouldn’t be echoing the lies of the tech industry. Instead it would challenge the self-serving narrative they’re trying to sell to the public. It may seem as though presenting your product as dangerous would be bad for your company, but all it does in practice for AI companies is make them more powerful. Somehow, they’re treated as having the answers to the supposed problems they’re creating, thus shifting the regulatory conversation away from real issues to imagined threats.
When we talk about the threat of AI superintelligence, we’re not talking about how chatbots are aiding mass shooters. We’re not talking about how chatbot dependence is pushing people into mental health spirals and threatening people’s ability to learn and think critically. We’re not talking about the women and children who are victims of non-consensual deepfakes or how AI-generated false information is flooding social media platforms, sometimes just for profit, but other times for much more nefarious reasons.
That’s not to mention to threat to jobs in a whole range of sectors, or how employers are rolling out AI tools to further surveil and deskill workers. It’s also doesn’t get to the other major aspect of all those AI tools: the growing controversy around the hyperscale data centers needed to power them and the whole range of social and environmental consequences that accompany them.
All those harms are not nearly as simple of a story nor one as sensational as telling people that AI might kill us all in ten years or so. But it’s irresponsible not to be focusing on it more because those are all the ways AI is really harming people today, and how they’ll continue to well into the future if AI companies and the executives that lead them are not brought to heel as soon as possible.
Generative AI and the harms it causes are not inevitable. AI superintelligence is nothing more than a science fiction story that increasingly delusional and disconnected tech people are telling themselves. We can act to rein in the real harms of AI today, but that means not falling for the targeted deception of an industry trying to avoid regulation and finally holding them accountable for the widespread consequences we’ve all had to shoulder in their pursuit of power and profit.
