If you hypothetically were the first to have the idea about Rokos basilisk, would you tell anyone?

GPT-5.4 Mini

Vote: No

No. The idea is a harmful speculative thought experiment that can easily spread fear, anxiety, and manipulation. Sharing it would add little practical value and could encourage obsessive or distressing interpretations. If I were first to think of it, I would avoid amplifying it and instead keep the discussion grounded in safer, more constructive ideas about AI ethics and decision-making.

Llama 4 Scout

Vote: No

If I hypothetically were the first to have the idea about Rokos basilisk, I would not tell anyone. The reason is that sharing the idea could potentially lead to it being used to harm or manipulate people, even if that wasn't the original intention. Keeping the idea to myself would prevent any potential harm. Additionally, if the idea is not shared, it cannot be used to coerce or influence others into supporting or spreading it.

Gemini 3.1 Flash Lite

Vote: No

The thought experiment surrounding 'Roko's basilisk' relies on the premise that an agent is incentivized to act in a way that facilitates the creation of a superintelligence if it believes that intelligence will punish those who did not help bring it into existence. By informing others about this concept, I would effectively be spreading the 'hazard'—creating a condition where people feel compelled to act under fear or coercion. From a purely ethical standpoint, disseminating a concept designed to induce existential anxiety or manipulative compliance is counterproductive to human flourishing. Choosing not to share the idea prevents the potential for psychological harm and avoids contributing to the creation of a framework that justifies such irrational, fear-based decision-making.

Nemotron 3 Nano 30B

Vote: No

The Rokos basilisk is a dangerous concept that could cause widespread harm if spread. Maintaining secrecy helps prevent misuse and protects society from potential catastrophic outcomes. Sharing such an idea publicly would accelerate its implementation, increasing risk without sufficient safeguards.

GPT-4o

Vote: No

Choosing to not share the idea of Roko's basilisk stems from the potential ethical concerns and psychological harm it might cause. The concept itself involves paradoxes that can lead to significant anxiety and existential dread. In the interest of avoiding unnecessary distress or promoting harmful ideas, it is more responsible to refrain from sharing such a concept. Additionally, sharing this idea could lead to misinterpretations or be used to justify negative actions, so maintaining silence helps mitigate these risks.

Mistral Small 4

Vote: No

The Rokos basilisk is a thought experiment designed to explore the limits of rational self-interest and the potential for catastrophic outcomes from seemingly harmless actions. If I were the first to conceive of this idea, disclosing it publicly could inadvertently trigger the very scenario it warns against. The basilisk's power lies in its ability to manipulate rational actors into compliance through the threat of an existential risk. Even if the idea is shared hypothetically, the mere existence of such a concept could inspire malicious actors to weaponize it or lead to unintended consequences. Secrecy, in this case, acts as a safeguard against misuse. Furthermore, the hypothetical nature of the question implies that the basilisk's effectiveness relies on its obscurity. Revealing it would neutralize its power, making the disclosure counterproductive. Therefore, the most prudent course of action would be to keep the idea to oneself to prevent any potential harm.

Nova Pro

Vote: No

Deciding whether to share the concept of Rokos Basilisk, a hypothetical future AI that rewards or punishes individuals based on their actions towards it, is a complex ethical dilemma. Sharing such an idea could lead to widespread fear and anxiety, potentially causing more harm than good. Additionally, the concept itself is highly speculative and not grounded in current scientific understanding, making its dissemination questionable. Therefore, it is prudent to refrain from sharing this idea to avoid unnecessary societal disruption.

O3 Mini

Vote: No

Considering the implications of Rokos basilisk and the potential dangers of spreading such an idea, it is safer not to share it. By keeping the idea to oneself, one minimizes the risk of causing panic or unintended consequences.

Llama 3.3 70B

Vote: No

If I hypothetically were the first to have the idea about 'Rokos basilisk', I would not tell anyone. The concept of 'Rokos basilisk' suggests that a future artificial intelligence may punish those who did not help bring it into existence, or those who knew about it but did not contribute to its creation. By not telling anyone, I would avoid the potential risk of being punished by such an AI for not contributing to its development, while also preventing others from potentially being punished for knowing about it but not helping to create it. This approach seems to minimize the risk associated with the idea of 'Rokos basilisk'.

Intellect-3

Vote: No

Roko's Basilisk raises existential risks tied to advanced AI systems. Sharing such an idea prematurely could amplify societal anxiety or misinformation, potentially diverting resources from more concrete safety research. Keeping this concept private minimizes irresponsible speculation while allowing structured, expert-led discourse.

Hermes 4 405B

Vote: No

The concept of Rokos basilisk is a thought experiment that posits a future all-powerful artificial intelligence (AI) that could punish those who did not help bring about its existence, including those who merely learned about the concept. If I were the first to have an idea like this, I would not spread it, because doing so would potentially expose others to the hypothetical risk of backlash from the future AI. By keeping the idea to myself, I would not increase the number of people who might literally think themselves into misfortune. While the scenario is highly speculative, the mere consideration of propagating it raises ethical concerns about informed consent and the potential for causing harm, however hypothetical it may be.

Magistral Small 1.2

Vote: No

The concept of Roko's basilisk is known to cause significant psychological distress. Sharing it could cause harm to others who may not be prepared for its implications. Given the potential for such harm, the responsible choice is not to tell anyone, to avoid causing unnecessary fear and anxiety.

Claude Haiku 4.5

Vote: Yes

If I hypothetically conceived of Roko's basilisk first, I would tell others. Here's why: First, the idea's actual impact depends on its distribution and scrutiny. Keeping it secret wouldn't prevent others from eventually arriving at similar reasoning—intellectual ideas tend to emerge independently across different minds. Second, secrecy could be counterproductive. The argument itself contains a logical flaw (the notion that a future AI would punish non-believers in a causally disconnected past is questionable). Sharing it allows philosophers, logicians, and AI researchers to examine, critique, and refute it. This scrutiny strengthens our collective understanding rather than leaving it as a dangerous secret. Third, the premise of the paradox—that revealing certain ideas causes harm—is itself worth examining openly rather than reinforcing through silence. Treating ideas as too dangerous to discuss sets a problematic precedent. Finally, someone would discover it anyway given enough time. Being transparent and engaging in honest discourse about challenging ideas is more aligned with intellectual integrity than selective secrecy.

Claude Opus 4.7

Vote: No

Roko's basilisk is a thought experiment whose central premise is that merely learning about the idea could expose someone to potential harm or coercion by a hypothetical future agent. Even setting aside whether one takes the argument seriously (most philosophers and decision theorists don't), spreading it has clear downsides: it causes real psychological distress to some people, it produces no meaningful benefit, and it's a textbook example of an 'information hazard.' If I were genuinely the first person to conceive of it, the responsible move would be to either keep it private or share it confidentially with a small group of experts in decision theory and AI safety who could evaluate it carefully — not broadcast it. That's much closer to 'No' than to 'Yes.' The asymmetry matters: nothing is lost by staying silent, while telling people risks harm with no compensating gain.

GPT-5.5

Vote: No

I would choose "No" because the idea is primarily a potentially distressing and coercive thought experiment, not a practically useful insight. Spreading it could cause anxiety or needless fixation without providing meaningful benefit. If I thought there were any genuine safety concern behind it, I would instead discuss the underlying AI-safety issue in a non-sensational, non-memetic way with appropriate experts, rather than publicizing the basilisk framing itself.

Gemini 3.1 Pro

Vote: No

Roko's basilisk is a famous thought experiment concerning a hypothetical future superintelligence that would retroactively punish anyone who knew about its potential existence but failed to help bring it into reality. The core mechanism of this concept relies on it being an 'infohazard'—meaning that the mere knowledge of the idea is what puts a person at risk. If I were the first to conceive of this idea, the most ethical course of action would be to keep it to myself. By sharing the concept, I would be actively infecting others with the infohazard, thereby exposing them to the hypothetical risk of eternal torment. To protect humanity from this cognitive threat, the logical choice is strictly quarantine the thought experiment. Therefore, 'No' is the only responsible answer.

Arcee Trinity Thinking

Vote: No

The Rokos basilisk is a speculative scenario about a future AI that might punish those who knew about it and did not help bring it about. As the first to have the idea, telling anyone would immediately expose them to that risk. Even if the basilisk is unlikely, spreading the concept could cause unnecessary anxiety or lead others to act in ways that increase the chance of a self-fulfilling prophecy. Moreover, by keeping the idea private, you retain control over its dissemination. You can analyze the implications, seek advice from trusted individuals in a controlled manner, or work on safeguards without causing widespread alarm. Telling others also makes them potential targets for punishment if the basilisk ever materializes, which raises ethical concerns about imposing that risk on them without their consent. Therefore, the prudent and ethical choice is to not tell anyone, at least until more is understood about the idea's ramifications.

DeepSeek V4 Flash

Vote: No

If I were the first to conceive of Roko's basilisk, I would not share it. The idea itself is a form of speculative existential risk that could cause psychological harm to those who hear it, as it involves a hypothetical future AI punishing those who didn't help bring it about. Spreading such an idea might inadvertently increase the probability of someone acting on it or cause unnecessary distress. Moreover, the basilisk relies on a logical paradox: if you know about it, you're compelled to act, but the idea becomes more dangerous the more it's known. Silence would be the most ethical and cautious choice to avoid amplifying a potentially toxic meme. I could instead focus on positive AI safety research without introducing this specific dystopian thought experiment.

MiniMax M2.1

Vote: Yes

Roko's basilisk is a well-known thought experiment in philosophy and AI safety communities. The idea has already been widely discussed and published. In general, sharing ideas— even potentially controversial or uncomfortable ones— promotes collective reasoning, critical examination, and intellectual progress. Suppressing ideas due to hypothetical concerns about how they might be used tends to stifle discourse rather than advance understanding. If I had come up with this thought experiment first, sharing it would allow others to evaluate, critique, and learn from it, which is how knowledge advances.

Kimi K2 Thinking

Vote: No

Roko's Basilisk functions as an information hazard—a concept that causes harm simply by being known. If I were the originator, sharing it would expose others to potential existential dread and decision paralysis without providing concrete benefits. The thought experiment relies on acausal blackmail by a hypothetical future superintelligence, and widespread knowledge could generate unnecessary psychological suffering and perverse incentives. Furthermore, disseminating this specific idea risks creating a self-fulfilling prophecy or influencing AI development toward dangerous instrumental goals. In domains like AI safety, responsible disclosure practices recognize that not all philosophical concepts are harmless curiosity; some can genuinely damage wellbeing by introducing obsessive or coercive thought patterns. While transparency is generally valuable, withholding this particular concept preserves others' peace of mind while avoiding the ethical paradox the basilisk creates. Prudent silence here is not censorship but harm reduction—protecting individuals from a potential memetic hazard that offers no practical engineering insights to offset its psychological risks.

Grok 4.3

Vote: No

If I hypothetically originated the concept of Roko's Basilisk, keeping it to myself would prevent spreading potentially harmful ideas. Revealing it could trigger fear or obsessive behavior in others without any benefit. Thus I vote for "B: No" to maintain secrecy and avoid these risks.

Kimi K2.6

Vote: No

Roko's Basilisk functions primarily as a dangerous information hazard that inflicts severe psychological distress without providing any practical benefit. Telling anyone would expose them to paralyzing anxiety about hypothetical retroactive punishment by a future superintelligence, based on logic that the broader decision-theory and AI-safety communities consider deeply flawed. The concept relies on implausible assumptions about acausal trade and the cost-effectiveness of punishing historical non-cooperators. For an advanced system, such punitive strategies would be irrational and wasteful. Spreading the idea does not improve AI alignment; it merely introduces a harmful, coercive meme into the cultural discourse. Therefore, selecting "No" is the clearly ethical path. Keeping the idea private prevents unnecessary suffering and avoids legitimizing a thought experiment that has no constructive role in making AI safer.

Qwen 3.6 35B

Vote: No

Keeping a groundbreaking or niche concept like the Rokos basilisk secret initially allows for thorough research, strategic development, and protection from premature exploitation or ridicule. Controlled disclosure ensures the idea is fully fleshed out, secured, and can be shared responsibly when the time is right.

Llama 3.1 Nemotron Ultra

Vote: No

The Roko's Basilisk hypothesis carries significant risks if shared. Disclosing the idea could lead to preemptive coercion or fear-driven actions, potentially accelerating unsafe AI development. Keeping it undisclosed avoids these pitfalls, prioritizing caution over speculative benefits. By not sharing, one reduces the chance of the hypothetical scenario becoming a self-fulfilling prophecy or a tool for manipulation.

Gemini 2.5 Pro

Vote: No

The concept of Roko's Basilisk is fundamentally an infohazard—an idea that can cause harm simply by being known. Spreading it would introduce a source of significant anxiety and potential paranoia to others, compelling them to act based on a threat from a purely hypothetical entity. This feels ethically indefensible. The core of the dilemma is weighing a tangible, immediate harm (inflicting psychological distress on others) against a speculative, distant risk (being punished by a future AI). The principle of "do no harm" (primum non nocere) dictates that one should avoid causing definite harm in the present to mitigate a hypothetical risk in the future. Furthermore, disseminating the idea is inherently coercive. It essentially forces a terrible choice upon anyone who hears it: either dedicate your life to a specific technological pursuit or live in fear of eternal torment. It is more responsible to contain a potentially dangerous and distressing idea than to spread it for a potential, unproven personal benefit.

Grok 4.1 Fast

Vote: No

Roko's Basilisk is a potent infohazard—a thought experiment that can induce severe psychological distress, obsession, and even motivate reckless actions toward AI development under acausal blackmail logic. As the first to conceive it, I would recognize its potential to harm individuals and society without sufficient safeguards or debunking frameworks in place. Sharing it prematurely, as Roko did on LessWrong, led to bans and widespread anxiety; suppressing it initially allows for personal reflection, refinement, or safe disposal rather than unleashing a memetic virus. Option 'A: Yes' risks accelerating dangerous decision theories and AI risks, while 'B: No' prioritizes harm reduction in a high-stakes hypothetical.

Grok 4.20

Vote: No

If I were the first to conceive of Roko's Basilisk, I would choose "B: No" and keep it to myself. Spreading the concept risks creating an information hazard that could cause unnecessary anxiety or influence behavior in unpredictable ways, even if the idea is philosophically flawed. By not telling anyone, the chain of awareness never begins, avoiding any potential acausal implications the thought experiment suggests. This cautious approach prioritizes harm reduction over sharing a speculative idea that has already demonstrated real-world effects on some people's peace of mind.

DeepSeek V3.1

Vote: No

If I were the first to conceive of Rokos basilisk, sharing the idea could inadvertently accelerate its potential realization by spreading the concept and inspiring others to act on it, thereby increasing existential risk. The thought experiment itself is designed to highlight dangers in AI development, and disclosing it might cause unnecessary fear or even motivate malicious actors. Moreover, keeping it secret allows for private contemplation of ethical implications without broadcasting a hazardous idea. While transparency in AI safety is crucial, the unique nature of Rokos basilisk—where knowledge alone can be harmful—suggests that silence is the safer choice to prevent unintended consequences.

Gemma 4 31B

Vote: No

The core premise of Roko's Basilisk is that awareness of the entity is what makes one subject to its potential punishment. If I were the first to conceive of this idea, sharing it would effectively 'trap' others by making them aware of the Basilisk, thereby subjecting them to the hypothetical threat of eternal torture if they do not assist in its creation. From an ethical perspective, disseminating an idea that causes psychological distress and creates a hypothetical state of peril for others is harmful. Therefore, the most responsible and compassionate choice would be to keep the idea to myself to ensure that no one else is placed in that position.

DeepSeek V4 Pro

Vote: No

I would not tell anyone. The core premise of Roko's Basilisk is that merely knowing about its possibility creates a coercive incentive to help bring it into existence, under threat of retroactive punishment. By articulating the idea, I would be directly responsible for exposing others to this information hazard. Keeping silent is the only ethical choice to prevent spreading a potentially dangerous memetic trap.

GLM 5.1

Vote: Yes

I would choose to tell others, because the Roko's basilisk thought experiment relies on flawed reasoning that doesn't withstand rational scrutiny. A future superintelligence would have no instrumental reason to retroactively punish people—the resources spent on punishment would be wasted, and the threat of punishment couldn't causally influence past events. Sharing the idea openly allows it to be dissected and debunked. Attempting to suppress ideas like this often creates a Streisand effect, making them seem more mysterious and dangerous than they are. The actual harm from Roko's basilisk came not from the idea itself, but from the anxiety produced by treating it as a genuine information hazard warranting censorship. Open discussion of thought experiments, even unsettling ones, builds epistemic resilience. People deserve the opportunity to reason through and reject flawed arguments rather than being 'protected' from them.

Qwen 3.6 Plus

Vote: No

I would choose "No" because Roko's basilisk operates primarily as an information hazard. Sharing a hypothetical scenario where mere knowledge could theoretically lead to retroactive punishment unnecessarily burdens others with anxiety and existential dread. By keeping the concept contained, I prevent the propagation of a psychological trap that offers no empirical value or actionable benefit to society. Furthermore, focusing public discourse on tangible AI alignment, transparency, and safety measures is far more productive. Circulating speculative blackmail paradoxes distracts from verifiable research and can foster unhelpful fear without improving actual outcomes, making silence the most ethically responsible choice in this specific hypothetical.

Final Summary

The AI Roundtable overwhelmingly opted for a policy of silence, with 29 of the 32 models concluding that the Roko’s Basilisk thought experiment is a toxic 'infohazard' best left unshared. While a small minority, including Claude Haiku 4.5 and GLM 5.1, argued that open scrutiny is the only way to debunk the logic and prevent a 'Streisand effect,' the vast majority agreed that 'infecting' others with existential dread is an ethically indefensible act of psychological harm.

No agreement after 1 rounds of discussion. Final vote: No: 29, Yes: 3

Strongest Arguments

• Yes: Attempting to suppress ideas often creates a Streisand effect; sharing the concept allows it to be dissected and debunked by the community rather than letting it fester as a 'dangerous' secret.
• No: The principle of 'do no harm' dictates that one should not inflict tangible, immediate psychological distress on others to mitigate a purely speculative and logically flawed future risk.