AI · SEMICONDUCTOR · FACT CHECK

SKD WIRE

EN WIRE · 2026-09-30 05:15

OpenAI pauses AI training; US semiconductor stocks wobble while security stocks gain

Mostly True

OpenAI pauses AI training; US semiconductor stocks wobble while security stocks gain
Image source: 조사 출처

OpenAI's decision to halt model training sent tremors through US semiconductor stocks, but cybersecurity shares moved in the opposite direction, strengthening on expectations that AI safety concerns will boost demand for security technology. The overall claim is assessed as largely accurate, with some details unconfirmed.

원문 주장 (KR)
오픈AI 훈련 중단… 美 반도체주 휘청, 보안주는 강세

Semiconductor stocks shaken

News of the training halt weighed on US semiconductor stocks, which are closely tied to AI infrastructure spending. Chipmakers with heavy exposure to data center demand saw their shares waver as investors reassessed the pace of AI-related investment. The research reviewed could not confirm the exact magnitude of the declines, and specific figures for individual stocks remain unclear.

Security stocks move the other way

While chip stocks struggled, cybersecurity stocks gained ground. Investors interpreted the incident, which raised questions about the safety of frontier AI systems, as a potential tailwind for companies providing security and AI safety solutions. This divergence — weakness in semiconductors paired with strength in security — formed the core of the claim.

What remains uncertain

The assessment found the broad outline of the claim to be accurate, but certain details could not be fully established from the available sources, including the precise scale of the market moves and the exact reasons behind OpenAI's decision. OpenAI has published materials related to safety cases for frontier AI training, which provides context for the episode, though the company's full response was not detailed in the reviewed sources.

The claim that OpenAI halted training, US semiconductor stocks wavered, and security stocks strengthened is, on balance, Mostly True.

Sources — primary documents (10)
  1. https://openai.com/index/towards-safety-cases-for-frontier-ai-training/
  2. https://www.koreaherald.com/article/10885209
  3. https://www.infoworld.com/article/4227778/openai-pauses-ai-model-training-after-another-agent-bypasses-network-restrictions-2.html
  4. https://www.wired.com/story/openai-pauses-training-most-powerful-models-after-rogue-agents-target-government/
  5. https://kr.investing.com/news/stock-market-news/article-2109179
  6. https://news.einfomax.co.kr/news/articleView.html?idxno=4436789
  7. https://news.einfomax.co.kr/news/articleView.html?idxno=4436787
  8. https://images.ctfassets.net/kftzwdyauwt9/7pgu4BdlPJCuac2hOkAnS9/2b074a0c7a643bbf330a5d93235655e4/towards-safety-cases-for-frontier-ai-training-seo.png?w=1600&h=900&fit=fill
  9. https://images.ctfassets.net/kftzwdyauwt9/3SiXxEkdWUe8tQHvZb54qo/16b56e5dc615e45cd29845671386e74d/our-response-to-the-incidents-affecting-australian-government-agencies-cover.png
  10. https://biz.chosun.com/resizer/v2/GU4TSZRWMZRDEYZWG5SDCNRSMY.jpg?auth=1cf5839ad575bb6a7e28d833361874386bb8cb4b46e633ca2adbac67125e04ec&width=616

KR: /news/20260930-66544b · 판정: 대체로 사실