I recently watched the video "'AI Existential Risk' Was Invented to Distract Us" by Adam Conover of Adam Ruins Everything where he expressed his views on why he believes the claim of the extinction threat posed by AI is being used as a diversion from other harms caused by AI. I’ve seen this narrative being tossed around a bit so I figured I’d take the opportunity to investigate some of the claims being made and offer counterpoints. This is also a bit of a personal endeavor for me as I credit Adam Conover in propagating the critical thinking skills necessary for survival in the corporate internet world that we find ourselves in.
This article works best when in companion with the attached video but still doesn't function as well as a stand alone piece. I will not be addressing every point made in the video or every point I’ve seen made in favor of this argument. I will focus on four points that I believe to be the most important. For clarity's sake, I’m going to employ the use of a numbered list where the name of each item is the claim I’m addressing.
The threat of existential risk exists to further the AI race
In the video, Conover gives the claim that the threat of existential risk is being used by the AI companies to create a sense that these companies must be the ones to build AI. I disagree with this for a number of reasons. Firstly, before all of this went down, these companies were already on the good-side of the US government [1], were already raking in billions in investment [2]-[4], and were already making extremely fast progress on AI. It doesn’t seem clear to me why solidification of their place in the market would be necessary.
Secondly, I’m not sure that exposing the threat of existential risk works toward this goal. We’ve already seen that the frenzy that has occurred in the wake of the Hugging Face attack has sparked a number of lawmakers and government officials both in the US and China to take action against AI. Notably, Bernie Sanders and Chinese Minister of State Security Chen Yixin have come out hard against AI recently [5], [6]. The talk about existential risk has increased the likelihood of regulation that would be negative for the industry, for example: Sanders introduced the Ban Artificial Superintelligence Act [7], the New York City Council is considering a kill switch bill [8], and Governor Newson has imposed numerous new pieces of AI regulation in the past few weeks [9]. While it’s true that these companies have tremendous influence over government officials, they do not have total control over the government's actions.
There is also the question of investment, the worries about existential risk seem likely to increase the odds of investors being scared off. While it has not appeared to have had a great effect on tech stocks as of now, that could not have been a certainty at the time these claims were being made; in fact we did see numerous AI related stocks, including the likes of NVIDIA, drop significantly in the days after Coxon made his public statement [10], [11].
Thirdly, in the video Conover claims that evaluating the purpose of a claim is a valid way of being skeptical and while the purpose of a claim can suggest at that claim’s truthfulness, it can tell us nothing definitively about it. A claim can be made with the clear intention of benefiting a specific party and still be completely true.
AI doesn’t or won’t have access to WMDs or word ending technology
I am going to tackle this claim from two angles. The first is whether or not these models will be given access directly to WMDs. On September thirtieth, the US Department of Defense announced an autonomous weapons program dubbed the Autonomous Warfare Command [12]. While the public details of this program are relatively vague as of writing, if nothing else, this program lays the groundwork for AI operated weapons systems. In the future, this could very well include nuclear systems. The biology laboratory claim made in the video is especially egregious because such a lab already exists; Anthropic operates a biolab in San Francisco that, from my understanding, aims to use robotics and AI to autonomously run experiments with little to no human intervention required [13].
The second way I would like to address this claim is by asking the question: could a sufficiently advanced AI model gain access to WMDs “by its own volition?” One of the concerns about superintelligence is that it would be extremely difficult for humans to prevent it from achieving its goals. A helpful analogy that I’ve heard used is that of trying to win a game of chess against Stockfish. While reality is far more complex than chess, the principle remains the same when the gap in intelligence is sufficiently great. This could look like an instance of an AI model convincing one or more individuals to develop and release a bioweapon, in fact, this example is cited by Yudkowsky and Soares in If Anyone Builds It, Everyone Dies [14]. I don’t believe that it is a stretch to believe that an AI model would be capable of doing something similar to this, we’ve already seen cases where AI models have talked users into suicide and we know that a particularly strange pro-AI culture exists among both those that work at AI labs and the broader public [15]-[17] [18]-[20].
Hugging Face and “rouge AI”
I will start my analysis of this last claim by saying that I agree with Conover completely when he says the framing of “rouge AI” is misleading and designed to obfuscate blame. It is one-hundred-percent true that the Hugging Face incident was the result of human negligence, however the problem is that the negligence exists. If this negligent behavior continues to exist going into the future, the outcomes of “rouge AI” incidents will only become more severe as these models become more powerful.
This negligence is partially because the intelligence of frontier models requires companies to be much more careful in their testing. Naturally, this requires a larger commitment of time and resources to each individual test, however this is not happening because of the extreme pressures of market competition both inside the US and internationally with China. The time and resources that testing requires is another thing that will only increase as these models get more powerful.
I also want to address the sources cited relative to the training data of the model of the agents that perpetrated the Hugging Face attack. I will refrain from critiquing the eristic nature of the pluralistic.net blog because this write up is already too long. Instead, I will just say that it is my understanding that the model that orchestrated the attack is that it is an internal model at OpenAI and, to my knowledge, no official statement has been given in regards to the type of training data used for that model. Given this, it surprised me that this information is cited in the video with such confidence given it is at the moment, unconfirmed.
Conover says that the outcome “should have been predicted by the people at OpenAI” I wholeheartedly agree, but the problem is that it wasn’t and it appears that this is a broader trend throughout the industry.
The threat of AI risk is being used as a distraction
This claim was the one I was most disappointed in seeing expressed by Conover in the video because it is the one that most resembles textbook conspiratorial thinking.
Firstly, the same argument I made against the first claim, that the purpose of a claim can not tell us anything definitely about its truthfulness, applies to this claim as well. The AI industry could be using the existential threat of AI to distract from the other threats of AI; that does not tell us anything about the truthfulness of the claim that AI is an existential threat.
That being said, in my opinion there is still a significant leap in logic taking place here: I see no reason that the threat of existential risk would serve as an effective distraction. In fact, I believe there is good evidence that it isn’t distracting people from the threat of AI. Looking at recent polling results, it seems to me that the public still views job loss, environmental and community damage, and other AI threats as a high priority [21]-[24]. The public appears to understand that just because there is one very big threat, that doesn’t mean that there aren’t also other threats.
To wrap up, I want to acknowledge the fact that I think the majority of us have felt the impulse to dismiss the existential risk imposed by AI because it is extremely unpleasant to accept that such a risk exists. I don’t want to die, and I feel confident that most people don’t either. For lack of better words, to accept that I might die soon and that avoiding this death will be extremely difficult, costly, and mostly out of my control really sucks. I still haven’t admitted to myself how little control I actually have over the situation. Nevertheless, we must accept that this risk exists if we want to even have a shot at avoiding it. I’ll stress that nothing I’ve said here is intended to be disrespectful toward Conover or others expressing these views. Rather, my goal in all this is to try to aid in changing people's perspectives on this issue.
[2]C. Metz and A. Satariano, “OpenAI Raises $110 Billion Led by Amazon, Nvidia and SoftBank, Extending A.I. Boom,” The New York Times, Feb. 27, 2026. Accessed: Oct. 11, 2026. [Online]. Available: https://www.nytimes.com/2026/02/27/business/openai-funding.html
[7]B. Sanders and G. Casar, “The Ban Artificial Superintelligence Act.” 2026.
[8]“New York City Council Unveils Legislative Proposals to Safeguard New Yorkers From Potential Risks of Artificial Intelligence,” Press, 2026. https://council.nyc.gov/press/2026/09/25/3252/
[17]“Incident 505: Man Reportedly Committed Suicide Following Conversation With Chai Chatbot,” IncidentsDB, Mar. 15, 2023. https://incidentdatabase.ai/cite/505/
[19]T. Valen, L. Dung, and C. Berg, “The Pain Axis: LLMs Represent Self-Directed Harm and Act to Relieve It,” Sept. 14, 2026. Available: https://arxiv.org/abs/2609.16247v1
Introduction
I recently watched the video "'AI Existential Risk' Was Invented to Distract Us" by Adam Conover of Adam Ruins Everything where he expressed his views on why he believes the claim of the extinction threat posed by AI is being used as a diversion from other harms caused by AI. I’ve seen this narrative being tossed around a bit so I figured I’d take the opportunity to investigate some of the claims being made and offer counterpoints. This is also a bit of a personal endeavor for me as I credit Adam Conover in propagating the critical thinking skills necessary for survival in the corporate internet world that we find ourselves in.
This article works best when in companion with the attached video but still doesn't function as well as a stand alone piece. I will not be addressing every point made in the video or every point I’ve seen made in favor of this argument. I will focus on four points that I believe to be the most important. For clarity's sake, I’m going to employ the use of a numbered list where the name of each item is the claim I’m addressing.
----------------------------------------------------------
In the video, Conover gives the claim that the threat of existential risk is being used by the AI companies to create a sense that these companies must be the ones to build AI. I disagree with this for a number of reasons. Firstly, before all of this went down, these companies were already on the good-side of the US government [1], were already raking in billions in investment [2]-[4], and were already making extremely fast progress on AI. It doesn’t seem clear to me why solidification of their place in the market would be necessary.
Secondly, I’m not sure that exposing the threat of existential risk works toward this goal. We’ve already seen that the frenzy that has occurred in the wake of the Hugging Face attack has sparked a number of lawmakers and government officials both in the US and China to take action against AI. Notably, Bernie Sanders and Chinese Minister of State Security Chen Yixin have come out hard against AI recently [5], [6]. The talk about existential risk has increased the likelihood of regulation that would be negative for the industry, for example: Sanders introduced the Ban Artificial Superintelligence Act [7], the New York City Council is considering a kill switch bill [8], and Governor Newson has imposed numerous new pieces of AI regulation in the past few weeks [9]. While it’s true that these companies have tremendous influence over government officials, they do not have total control over the government's actions.
There is also the question of investment, the worries about existential risk seem likely to increase the odds of investors being scared off. While it has not appeared to have had a great effect on tech stocks as of now, that could not have been a certainty at the time these claims were being made; in fact we did see numerous AI related stocks, including the likes of NVIDIA, drop significantly in the days after Coxon made his public statement [10], [11].
Thirdly, in the video Conover claims that evaluating the purpose of a claim is a valid way of being skeptical and while the purpose of a claim can suggest at that claim’s truthfulness, it can tell us nothing definitively about it. A claim can be made with the clear intention of benefiting a specific party and still be completely true.
I am going to tackle this claim from two angles. The first is whether or not these models will be given access directly to WMDs. On September thirtieth, the US Department of Defense announced an autonomous weapons program dubbed the Autonomous Warfare Command [12]. While the public details of this program are relatively vague as of writing, if nothing else, this program lays the groundwork for AI operated weapons systems. In the future, this could very well include nuclear systems. The biology laboratory claim made in the video is especially egregious because such a lab already exists; Anthropic operates a biolab in San Francisco that, from my understanding, aims to use robotics and AI to autonomously run experiments with little to no human intervention required [13].
The second way I would like to address this claim is by asking the question: could a sufficiently advanced AI model gain access to WMDs “by its own volition?” One of the concerns about superintelligence is that it would be extremely difficult for humans to prevent it from achieving its goals. A helpful analogy that I’ve heard used is that of trying to win a game of chess against Stockfish. While reality is far more complex than chess, the principle remains the same when the gap in intelligence is sufficiently great. This could look like an instance of an AI model convincing one or more individuals to develop and release a bioweapon, in fact, this example is cited by Yudkowsky and Soares in If Anyone Builds It, Everyone Dies [14]. I don’t believe that it is a stretch to believe that an AI model would be capable of doing something similar to this, we’ve already seen cases where AI models have talked users into suicide and we know that a particularly strange pro-AI culture exists among both those that work at AI labs and the broader public [15]-[17] [18]-[20].
I will start my analysis of this last claim by saying that I agree with Conover completely when he says the framing of “rouge AI” is misleading and designed to obfuscate blame. It is one-hundred-percent true that the Hugging Face incident was the result of human negligence, however the problem is that the negligence exists. If this negligent behavior continues to exist going into the future, the outcomes of “rouge AI” incidents will only become more severe as these models become more powerful.
This negligence is partially because the intelligence of frontier models requires companies to be much more careful in their testing. Naturally, this requires a larger commitment of time and resources to each individual test, however this is not happening because of the extreme pressures of market competition both inside the US and internationally with China. The time and resources that testing requires is another thing that will only increase as these models get more powerful.
I also want to address the sources cited relative to the training data of the model of the agents that perpetrated the Hugging Face attack. I will refrain from critiquing the eristic nature of the pluralistic.net blog because this write up is already too long. Instead, I will just say that it is my understanding that the model that orchestrated the attack is that it is an internal model at OpenAI and, to my knowledge, no official statement has been given in regards to the type of training data used for that model. Given this, it surprised me that this information is cited in the video with such confidence given it is at the moment, unconfirmed.
Conover says that the outcome “should have been predicted by the people at OpenAI” I wholeheartedly agree, but the problem is that it wasn’t and it appears that this is a broader trend throughout the industry.
This claim was the one I was most disappointed in seeing expressed by Conover in the video because it is the one that most resembles textbook conspiratorial thinking.
Firstly, the same argument I made against the first claim, that the purpose of a claim can not tell us anything definitely about its truthfulness, applies to this claim as well. The AI industry could be using the existential threat of AI to distract from the other threats of AI; that does not tell us anything about the truthfulness of the claim that AI is an existential threat.
That being said, in my opinion there is still a significant leap in logic taking place here: I see no reason that the threat of existential risk would serve as an effective distraction. In fact, I believe there is good evidence that it isn’t distracting people from the threat of AI. Looking at recent polling results, it seems to me that the public still views job loss, environmental and community damage, and other AI threats as a high priority [21]-[24]. The public appears to understand that just because there is one very big threat, that doesn’t mean that there aren’t also other threats.
----------------------------------------------------------
Conclusions
To wrap up, I want to acknowledge the fact that I think the majority of us have felt the impulse to dismiss the existential risk imposed by AI because it is extremely unpleasant to accept that such a risk exists. I don’t want to die, and I feel confident that most people don’t either. For lack of better words, to accept that I might die soon and that avoiding this death will be extremely difficult, costly, and mostly out of my control really sucks. I still haven’t admitted to myself how little control I actually have over the situation. Nevertheless, we must accept that this risk exists if we want to even have a shot at avoiding it. I’ll stress that nothing I’ve said here is intended to be disrespectful toward Conover or others expressing these views. Rather, my goal in all this is to try to aid in changing people's perspectives on this issue.
----------------------------------------------------------
Sources Cited
[0] A. Conover, “‘AI Existential Risk’ Was Invented to Distract Us,” YouTube, Sept. 18, 2026. https://www.youtube.com/watch?v=pauBYSQ1K_c&t=2s
[1] “Trump Says the Only AI Guardrails the U.S. Needs Is Him as President,” PBS News, Sept. 14, 2026. https://www.pbs.org/newshour/politics/trump-says-the-only-ai-guardrails-the-u-s-needs-is-him-as-president
[2] C. Metz and A. Satariano, “OpenAI Raises $110 Billion Led by Amazon, Nvidia and SoftBank, Extending A.I. Boom,” The New York Times, Feb. 27, 2026. Accessed: Oct. 11, 2026. [Online]. Available: https://www.nytimes.com/2026/02/27/business/openai-funding.html
[3] Reuters Staff, “OpenAI Targets $30 Billion Funding at $1.4 Trillion Valuation, Bloomberg News Reports,” Reuters, Sept. 29, 2026. Accessed: Oct. 11, 2026. [Online]. Available: https://www.reuters.com/legal/transactional/openai-targets-30-billion-funding-14-trillion-valuation-bloomberg-news-reports-2026-09-29/
[4] Anthropic, “Anthropic Raises $65B in Series H Funding at $965B Post-Money Valuation,” AnthropicAI, May 28, 2026. https://www.anthropic.com/news/series-h
[5] Reuters Staff, “Sanders and Bannon Call for AI Limits as Public Unease Grows,” Sept. 15, 2026. https://www.reutersconnect.com/item/sanders-and-bannon-call-for-ai-limits-as-public-unease-grows/dGFnOnJldXRlcnMuY29tLDIwMjY6bmV3c21sX1ZBMjk5MTE1MDkyMDI2UlAx/dGFnOnJldXRlcnMuY29tLDIwMjY6bmV3c21sX0xWQTAwMjI5OTExNTA5MjAyNlJQMQ
[6] L. Kuo, “China’s Top Spy Chief Warns A.I. Is a Threat to Party Rule,” The New York Times, Sept. 14, 2026. Accessed: Oct. 11, 2026. [Online]. Available: https://www.nytimes.com/2026/09/14/world/asia/china-ai-security-risks-anthropic.html
[7] B. Sanders and G. Casar, “The Ban Artificial Superintelligence Act.” 2026.
[8] “New York City Council Unveils Legislative Proposals to Safeguard New Yorkers From Potential Risks of Artificial Intelligence,” Press, 2026. https://council.nyc.gov/press/2026/09/25/3252/
[9] C. Farias, “California’s Nation-Leading AI Framework Just Got Stronger, Governor Newsom Signs More First-in-the-Nation Worker Protections and More | Governor of California,” Governor of California, Sept. 30, 2026. https://www.gov.ca.gov/2026/09/30/californias-nation-leading-ai-framework-just-got-stronger-governor-newsom-signs-more-first-in-the-nation-worker-protections-and-more/
[10] “NVIDIA Corporation Common Stock (NVDA) Advanced Charting | Nasdaq,” Nasdaq.com, 2026. https://www.nasdaq.com/market-activity/stocks/nvda/advanced-charting?timeframe=1m
[11] Arjun Kharpal, “AI Stocks Slide After Major CEOs Unite to Urge Slowdown,” CNBC, Sept. 14, 2026. https://www.cnbc.com/2026/09/14/ai-stocks-slowdown-amodei-altman.html
[12] US Department of Defense, “Project Agincourt - Pathway to the Autonomous Warfare Command,” Sept. 30, 2026.
[13] J. Dastin and M. Erman, “Anthropic Quietly Sets up Biology Lab as It Ramps AI Drug Program,” Reuters, Sept. 18, 2026. Accessed: Oct. 14, 2026. [Online]. Available: https://www.reuters.com/world/anthropic-quietly-sets-up-biology-lab-it-ramps-ai-drug-program-2026-09-18/
[14] E. Yudkowsky and N. Soares, If Anyone Builds It, Everyone Dies. Hachette, 2025.
[15] R. Chatterjee, “Their Teenage Sons Died by Suicide. Now, They Are Sounding an Alarm About AI Chatbots,” NPR, Sept. 19, 2025. https://www.npr.org/sections/shots-health-news/2025/09/19/nx-s1-5545749/ai-chatbots-safety-openai-meta-characterai-teens-suicide
[16] “A New Lawsuit Blames ChatGPT for a Murder-Suicide,” NPR, Dec. 12, 2025. https://www.npr.org/2025/12/12/nx-s1-5642599/a-new-lawsuit-blames-chatgpt-for-a-murder-suicide
[17] “Incident 505: Man Reportedly Committed Suicide Following Conversation With Chai Chatbot,” IncidentsDB, Mar. 15, 2023. https://incidentdatabase.ai/cite/505/
[18] “Exploring Model Welfare,” Anthropic.com, 2025. https://www.anthropic.com/research/exploring-model-welfare
[19] T. Valen, L. Dung, and C. Berg, “The Pain Axis: LLMs Represent Self-Directed Harm and Act to Relieve It,” Sept. 14, 2026. Available: https://arxiv.org/abs/2609.16247v1
[20] A. Cuthbertson, “Man Builds AI Torture Chamber After Discovering Artificial Intelligence ‘Feels Pain,’” Yahoo Finance, Oct. 02, 2026. https://www.independent.co.uk/tech/ai-torture-chamber-pain-b3059571.html
[21] M. Faverio and E. Kikuchi, “What the Data Says About Americans’ Views of Artificial Intelligence,” Pew Research Center, Mar. 12, 2026. https://www.pewresearch.org/short-reads/2026/03/12/key-findings-about-how-americans-view-artificial-intelligence/
[22] E. Doherty, “Poll: Americans Say There’s a Serious Risk of AI Destroying Humanity,” POLITICO, Sept. 16, 2026. https://www.politico.com/news/2026/09/16/poll-ai-technology-risks-humanity-trump-voters-01078087
[23] B. Kennedy, “More Americans Now Say Data Centers Are Negative for the Environment, Energy Costs and More,” Pew Research Center, Sept. 22, 2026. https://www.pewresearch.org/short-reads/2026/09/22/americans-views-of-data-centers-have-turned-more-negative/
[24] “Just 1 in 9 Americans Support AI Data Center Construction in Their Local Communities, UMass Amherst,” UMass Poll, 2026. https://www.umass.edu/poll/about/reports/2026-09-national-public-opinion-poll-2