오픈AI CEO 샘 알트만의 자택에 화염병을 투척한 20세 용의자가 체포되었습니다. 그는 AI가 인류를 멸망시킬 수 있다는 극단적인 우려를 품고 있었으며, AI 개발 중단을 주장하는 'PauseAI' 추종자였던 것으로 파악되었습니다. 이 사건은 AI 안전성을 둘러싼 사회적 긴장이 폭력적인 테러로 이어질 수 있음을 보여주는 중요한 사례입니다.
번역된 본문
2026년 4월 12일 두 번째 업데이트: AI 개발 중단을 옹호하는 단체인 PauseAI는 이번 공격을 강하게 규탄했습니다. 이 단체는 용의자인 모레노-가마가 PauseAI 활동가가 아니었다고 밝혔습니다. 이 단체는 성명을 통해 "우리는 샘 알트만, 그의 가족, 그리고 이 사건으로 피해를 입은 모든 이들의 안전과 평안을 기원합니다. PauseAI는 모든 사람이 안전할 자격이 있다고 믿기 때문에 존재하며, 여기에는 샘 알트먼과 그의 사랑하는 사람들도 포함됩니다. 누구에 대한 폭력도 우리가 추구하는 모든 것에 반하는 것"이라고 밝혔습니다.
PauseAI에 따르면, 이 남성은 약 2년 전 이 단체의 공개 디스코드 서버에 가입하여 34개의 메시지를 게시했으며 조직과 다른 관련은 없었습니다. 운영진은 그의 메시지 중 하나를 모호한 내용으로 간주해 경고를 발령한 바 있습니다. 공격이 발생한 후 그의 계정은 정지되었습니다. 운영진 중 한 명이 기본적인 차단 절차의 일환으로 그의 메시지를 삭제하기 시작했지만, 잠재적인 수사에 중요한 자료가 될 수 있다는 사실을 깨닫고 삭제를 중단했습니다.
PauseAI는 청원 및 대중 교육 캠페인과 같은 비폭력적인 형태의 항의를 엄격하게 추구한다고 밝혔습니다. 이 단체는 정당한 우려를 가진 사람들을 위한 평화적인 채널이 없다면 개인이 어떠한 지지 네트워크 없이 단독으로 행동할 위험이 커지고 있다고 경고합니다.
2026년 4월 12일 업데이트:
샘 알트만의 자택을 공격한 화염병 용의자는 AI로 인한 인류 멸망을 우려한 PauseAI 추종자였던 것으로 보입니다.
모레노-가마는 고도화된 AI 개발에 항의하고 이를 인류에 대한 잠재적 위협으로 간주하는 'PauseAI' 단체의 추종자였던 것이 분명합니다. 그는 자신의 서브스택(Substack)에 1월부터 3월 사이에 6편의 장문의 글을 게재했습니다. '인간을 위한 추도사(A Eulogy for Man)'라는 제목의 글에서 그는 AI에 의한 인류의 멸종에 대해 경고했습니다.
"인류 역사 속에서 더 발전된 인간 문명이 덜 발전된 문명과 접촉했을 때, 덜 발전된 집단은 정복과 제노사이드(학살)를 겪는 경우가 많았습니다. 그렇다면 우리는 왜 의도적으로 이런 일을 저지르려는 것입니까? 놀라운 질문입니다. 제 대답은 저도 모르겠다는 것입니다. 하지만 우리는 어쨌든 그것을 하고 있습니다. 최상위에 있는 몇몇 불안정한 남성들이 그렇게 결정했기 때문입니다. (이왕이면 이 과정에서 최소한의 안전장치를 두고 최대한 빨리 진행하고 있다는 점도 덧붙이겠습니다. 이게 마음이 편해진다면 말이죠.) 따라서 그 선택이 이미 우리를 대신해 이루어졌으므로, 다음 질문은 우리가 그저 누워서 이 일이 전개되도록 허용할 것인가 입니다."
"[...] 그래서 저는 마지막 두 가지 원형을 소개합니다. 전사(Warrior)와 순교자(Martyr)입니다. 인간의 자연스러운 일탈은 자신과 관련 없는 사람들을 위해 싸우며, 더 나아가 더 미친 무언가, 즉 '이념(Ideas)'을 위해 죽기도 할 것입니다. [...] 전사는 자신의 이념을 위해 죽을 준비가 되어 있을 뿐만 아니라, 죽일 준비도 되어 있습니다. 그는 전쟁에서 싸움으로써 자신의 이념을 무기로 구현합니다. 왜냐하면 세상이 자신의 행동으로부터 이익을 얻을 것이라고 믿기 때문입니다. [...] 비록 이것은 엄숙한 이야기이지만, 저는 우리가 이 원형들의 정신을 따를 것을 요청합니다. 스스로를 방어할 기회가 없는 사람들을 생각해야 하기 때문입니다. 읽어주셔서 감사합니다. 다음 생에서 다른 눈빛으로 다시 뵙기를 바랍니다. 우리는 싸울 가치가 있습니다." - '인간을 위한 추도사' 중 모레노-가마
통신 연구자 니리트 와이스-블랫(Nirit Weiss-Blatt)에 따르면, 그는 디스코드에서 '버틀러리안 지하디스트(Butlerian Jihadist)'라는 이름을 사용했는데, 이는 SF 소설 시리즈인 '듄(Dune)'에서 인류가 기계를 파괴하는 것을 참조한 것입니다. 12월 초, 그는 PauseAI 서버에 "우리는 자정에 가까워지고 있으며, 이제 실제로 행동할 때입니다"라고 적었습니다. 운영진은 폭력적인 행동을 촉구하는 글을 올리면 정지 조치할 것이라고 경고했습니다. PauseAI는 이후 해당 사용자를 디스코드 서버에서 퇴장시켰습니다.
업데이트: 샌프란시스코 스탠다드(The San Francisco Standard)에 따르면 용의자가 체포되었습니다. 20세의 다니엘 알레한드로 모레노-가마(Daniel Alejandro Moreno-Gama)는 2026년 4월 10일 금요일 오후 샌프란시스코 카운티 교도소에 수감되었습니다. 모레노-가마는 샘 알트만의 자택 금속 문을 향해 화염병을 던진 혐의를 받고 있습니다.
Man who firebombed Sam Altman's home was likely driven by AI extinction fears Matthias Bastian View the LinkedIn Profile of Matthias Bastian Apr 12, 2026 Nano Banana Pro prompted by THE DECODER Ask about this article… Search Second update, April 12, 2026: The organization PauseAI, which advocates for halting AI development, sharply condemned the attack . Moreno-Gama was not a PauseAI activist, the group said. "We wish safety and peace to Sam Altman, his family, and everyone affected. PauseAI exists because we believe everyone deserves to be safe, including Sam Altman and his loved ones. Violence against anyone is antithetical to everything we stand for," the organization wrote in a statement . Ad According to PauseAI, the man joined the group's public Discord server roughly two years ago, posted 34 messages, and had no other involvement with the organization. Moderators flagged one of his messages as ambiguous, which earned him a warning. After the attack, his account was banned. A moderator started deleting his messages as part of the standard ban process but stopped after realizing they could be relevant to a potential investigation. Ad DEC_D_Incontent-1 PauseAI says it strictly pursues nonviolent forms of protest, including petitions and public education campaigns. The organization warns that without peaceful channels for people with legitimate concerns, there's a growing risk of individuals acting alone and without any support network. Update from April 12, 2026: Ad Molotov suspect who attacked Sam Altman's home was likely a Pause AI follower with AI extinction fears Moreno-Gama was apparently a follower of the "Pause AI" group , which protests the development of advanced AI and considers it a potential threat to humanity. On his Substack , he published six lengthy posts between January and March. In one piece titled "A Eulogy for Man," he warned about the extinction of humanity through AI . Even within human history, whenever a more advanced human civilization has made contact with a less advanced one, the less advanced group is often met by conquest and genocide. So why the hell would we knowingly do this? Wonderful question, my answer is I have no idea, but we’re doing it anyway; because a few insecure men at the top decided so (we’re also doing it with as little guardrails as possible and also as fast as possible if that helps you sleep better.) So since that choice has already been made for us, the next question is do we just lay back and allow this thing to unfold? Ad DEC_D_Incontent-2 [...] Ad Thus this brings me to my two final archetypes: The Warrior & The Martyr As so the natural aberration of man would fight for those who he is not related to, so would it also die for something even more insane, it’s Ideas. [...] The Warrior is not only willing to die for his Ideas, but willing to kill too. He puts his ideas to metal by fighting in wars, because he believes the world will benefit from his actions. [...] Although this is a somber note I ask that we follow the spirit of these archetypes, as we must think of those who have no chance in defending themselves. Thank you for reading. I hope to see you again with different eyes in a different life. We are worth fighting for. Moreno-Gama in "A Eulogy for Man" On Discord, he used the name "Butlerian Jihadist," according to communications researcher Nirit Weiss- Blatt—a reference to the science fiction series "Dune," in which humanity destroys machines. In early December, he wrote on the PauseAI server: "We are close to midnight, it's time to actually act." A moderator warned him that calls for violence would result in a ban. Pause AI has since removed the user from its Discord server. Update: A suspect has been arrested, according to the San Francisco Standard . Daniel Alejandro Moreno-Gama, 20, was booked into the San Francisco County Jail on Friday afternoon, April 10, 2026. Moreno-Gama allegedly threw a Molotov cocktail at the metal gate of Sam Altman's home in the Russian Hill neighborhood around 3:40 a.m. Security personnel on site put out the fire, and surveillance cameras captured the whole thing. Shortly after, someone matching the suspect's description showed up outside OpenAI's headquarters in Mission Bay and threatened to burn the building down, according to the company. Police arrested the man at the scene. He faces charges including attempted murder, arson, possession, or manufacture of an incendiary device, and additional offenses. Original article: Someone threw a Molotov cocktail at OpenAI CEO Sam Altman's home in the middle of the night Someone threw a Molotov cocktail at OpenAI CEO Sam Altman's home at 3:45 a.m. In response, Altman published a personal blog post admitting past mistakes and comparing the AI industry's power struggles to the "Ring of Power." According to Altman, the Molotov cocktail bounced off the house, and nobody was hurt. He believes the incident is tied to a recently published critical profile of him, one he initially called "incendiary" and didn't take seriously enough. Altman, who has raised billions of dollars largely on the strength of his rhetoric and who is the driving force behind the "AI will change everything" narrative, now admits he's been "thinking that I have underestimated the power of words and narratives." Altman used the post to lay out his broader views on AI, many of which he has stated before: the technology needs to be democratized, and no small group of companies should control it. People's fear of AI is valid, he wrote, and society may be going through the biggest shift in a long time, perhaps the biggest ever. He also sees an urgent need for a "society-wide response" to the threats AI poses, he argued, including new policies to manage what he expects will be a "difficult economic transition." Altman also acknowledged mistakes, including being "conflict-averse, which has caused great pain for me and OpenAI," and mishandling the former OpenAI board situation . He recognized that OpenAI is now a major platform, not a startup, and needs to "operate in a more predictable way." Altman says he's proud of resisting Elon Musk's push for one-sided control over OpenAI and that, against all odds, they built powerful AI, raised the capital for infrastructure, and shipped secure services at scale. Many companies claim to change the world; OpenAI actually did, according to Altman. One AGI to… Altman compared the AI industry's internal conflicts to a "ring of power" dynamic: once you've seen AGI, you can't stop trying to control it. The only solution I can come up with is to orient towards sharing the technology with people broadly, and for no one to have the ring. The two obvious ways to do this are individual empowerment and making sure democratic system stays in control. Sam Altman OpenAI wants to "be a voice and a stakeholder," he said, "but not to have all the power." According to Altman, OpenAI plans to work within the democratic process, even when it's "messy and slower than we'd like." He called on the industry to tone down the rhetoric and "have fewer explosions in fewer homes, figuratively and literally." The call comes not only after the attack on his home, but also as tensions between OpenAI, Anthropic, and the Pentagon have recently escalated over military AI use . As for the New Yorker article that allegedly triggered the attack—the one Altman initially called "incendiary"—it was based on years of reporting and conversations with more than 100 people. Many of the most critical details came from Altman himself in direct exchanges with journalists, including his admission that his "vibes" didn't line up with traditional AI safety thinking . He later walked back the word "incendiary" on X , calling it a "bad word choice" after a "tough day." As of now, there's no information about who carried out the attack or why. AI News Without the Hype – Curated by Humans Subscribe to THE DECODER for ad-free reading, a weekly AI newsletter, our exclu