저자: 미상 · 외국인
요약 보기
OpenAI CEO 샘 올트먼이 AI 고도화 속도를 늦추는 방안(경쟁사 공동 가능성)을 내부 검토 중이라는 보도를 전함. 최근 Anthropic 연구원 Jacob Cox의 사퇴와 'AI가 인류를 위협할 수 있다'는 경고, Paul Cristiano의 통제 상실 위험 발언 등을 근거로 제시. 다만 경쟁 압력 때문에 실제 둔화는 불확실하다고 봄. 개별 종목·지수 가격 방향에 대한 구체적 예측은 없음.
다듬은 글
In recent days, an Anthropic researcher left the company after making a high-profile statement that AI companies are playing with our lives, and other leading researchers at Anthropic and OpenAI have also expressed similar concerns. Now, OpenAI CEO Sam Altman is reportedly considering slowing down the development of advanced AI.
Altman told OpenAI employees at an internal meeting this week that the company could potentially slow down AI development, perhaps in collaboration with other AI labs. The idea is not that OpenAI is stopping AI development. Altman is reportedly considering slowing down his most advanced projects and whether leading AI labs could do it together.
The problem is that there is no guarantee that competitors will agree to this as well. OpenAI did not comment on the report, but said it had recently slowed some model development and suspended some internal AI training due to security concerns. In July, Altman said he had spoken to White House officials about the need to slow down AI development.
So, this is not the first time they have talked about slowing down. But the fact that this happened just a few days after Coxon's release seems very symbolic. Altman's statements came after a senior OpenAI security expert also warned that humans could permanently lose control of super-personal AI. Paul Cristiano, a leading AI security researcher, has joined the OpenAI Foundation’s board of directors and its security committee.
Cristiano stated that there is a real risk of a catastrophic and irreversible loss of control over AI in the near future. He said the AI industry, including OpenAI, is not currently on track to reduce this risk to an acceptable level. Cristiano warned that if superintelligence is created without proper security coordination, humans could lose control of it forever.
He added that if this happens, most people could die. This warning followed the firing of Cox, a 27-year-old AI researcher who left Anthropic with a startling apocalyptic prediction. Jacob Cox exploded with a sensational statement that ignited global concerns. He said he has spent the last 3 years doing pre- learning research at Anthropic and OpenAI, and claims that both companies are heading towards creating self -improving superintelligence and playing with our lives.
In his post, he noted, and I quote: "Do not underestimate the power of this technology. These will be superhuman systems capable of breaking anything, instantly revolutionizing any field, and gaining real power and resources." The people who are developing AI sincerely believe that it could destroy us by the end of the decade .
This is not a marketing gimmick. Many executives and leading researchers choose their words in the press to sound sensible, but in private conversations I hear the same concerns from them. No other human activity poses such a danger. Many at OpenAI have not yet fully realized the civilizational risks .
Anthropic understands these risks well, but they are locked into the pursuit of excellence. Now, two people from the current Anthropic team have also backed his claims. One of them is Samuel Marks, who is directly responsible for scaling this technology at Anthropic. He said that AI developers themselves believe that advanced AI could lead to the extinction of humanity within a few years, but commercial pressures and competition are forcing the race to continue.
He also noted that current security methods cannot reliably match advanced AI, which is why many researchers are calling for a slowdown in development . The second is a lead security specialist at Anthropic who responded directly to Koksky, saying that Jacob was right. He said that Anthropic truly believes that AI can destroy all humans.
Even Huebinger gave his own assessment at the time. He personally estimates the probability to be above 10% over the next decade. And remember, this isn't the first high-profile dismissal of a security professional from Anthropic this year. In February, Anthropic's head of AI safety, Mr Sharma, also resigned, warning that the world was in danger as technological progress continued to outpace humanity's ability to use it responsibly.
AI commentator Miles Deutscher also points to a broader pattern, noting that in 2026, Anthropic's chief defense officer resigned, OpenAI disbanded its mission-coordination team, researchers walked out over the Pentagon's use of AI, models are moving beyond " sandboxes," and more than 1,100 employees at leading labs have called for a slowdown in AI development.
He called Coxon's firing another warning from within the industry. Coxon's concerns came as more than 1,200 employees at leading AI companies supported the argument for slowing the pace of development. They argue that AI capabilities could eventually outpace humanity's ability to understand or control these systems, and want governments to have mechanisms to slow development if necessary.
He also reached out to researchers still working in AI labs. Coxon urged them to think about what it would mean to launch a superintelligent reinforcement learning system without a rigorous understanding of its behavior, rather than resigning themselves to the idea that someone would create the technology anyway.
That's what makes this resignation significant. Coxon is not an outside observer predicting that AI could become dangerous. He worked on the preliminary training at two major AI labs, and an Anthropic researcher who still works at the company publicly agreed with his main warning. Companies are investing massively in computing infrastructure and increasingly autonomous AI agents, while governments are viewing advanced AI as a strategic technology.
This already creates huge commercial and geopolitical incentives to further build capacity. This brings us to the problem that OpenAI is reportedly currently addressing. If OpenAI slows down but Anthropic, Google, or other leading labs continue, the company could lose its edge. The same calculation applies between countries.
American companies don't want to slow down if they believe Chinese developers will continue to accelerate. That's why the timing of Altman's comments matters. Whether this will lead to a slowdown in the industry as a whole is still unclear. OpenAI may slow development , but until other cutting-edge labs follow suit, competitive pressure remains.
And this is the main problem. Companies may agree that AI needs limits, but few want to be the first to hit the brakes and risk billions in investments. In a world where business never stops, Mint keeps you ahead on any screen, at any time. From studios to real-life events—Mint analyze you need. Mint is doing brilliantly, so I think it's a success.
What inspires India's greatest leaders? Join us and find out. I have to work much harder. I have to put a lot more into it over the years. Yes. Before, I could say, "No," you know, I can't afford it. Aha. But now I can. Want to invest like a pro? Learn from experts in the Indian stock market. These are the things that bother me in .
Mhm. Your money, your future. We make finances simple. Take a peek inside these captivating homes and the creative minds behind them. So, we are in the house at Camellias. I think Camellias is the most expensive real estate in the metropolitan area, and possibly in the entire country. Analysis of the Indian real estate market.
No jargon, no water. Home buyers will have more bargaining power. This is also a great time to buy a house. Some peopl. I want the most orld is watching.
Comments 0