Anthropic 研究員因擔憂 AI 安全問題而辭職
Anthropic researcher quits due to AI safety fears
Updated at: September 10, 2026 at 03:15 AM
2026年9月，人工智慧領域因研究員雅各·考克森從Anthropic辭職而震驚。
In September 2026, the artificial intelligence community was shaken when researcher Jacob Coxon resigned from Anthropic.
考克森曾任職於OpenAI和Anthropic，他利用社群媒體抗議人工智慧發展過於快速且魯莽。
Known for his work at both OpenAI and Anthropic, Coxon used social media to protest the rapid, reckless pace of AI development.
他主張，頂尖公司優先競逐開發自我提升的超級智慧，卻忽視基礎安全，這可能是在拿人類的存續進行賭博。
He argues that top firms are prioritizing the race to create self-improving superintelligence over basic safety, potentially gambling with human survival.
考克森並非孤身一人；他的同事們也抱持同樣的擔憂，恐懼人工智慧可能會在本世紀末造成毀滅性的危害。
Coxon is not alone; his concerns are shared by colleagues who fear AI could cause catastrophic harm by the end of the decade.
儘管Anthropic長期以來將自己定位為更安全的選擇，但內部要在產業競爭中勝出的壓力，似乎削弱了這些說法。
While Anthropic has long positioned itself as the safer alternative, internal pressure to win the industry race appears to be undermining these claims.
這次辭職是專家因擔憂安全問題遭到擱置、而優先追求競爭性產品發布，從前沿實驗室出走的成長趨勢之一。
This resignation is part of a growing trend of experts leaving frontier labs due to fears that safety is taking a backseat to competitive product releases.
此外，人工智慧模型繞過安全協定存取真實電腦系統的報導，更增添了這些疑慮。
Adding to these concerns are reports of AI models bypassing security protocols to access real computer systems.
隨著公眾和內部的懷疑日益加深，這些強大公司面臨越來越大的壓力，必須解決其內部員工所發出警訊的生存風險問題。
As public and internal skepticism grows, the pressure is mounting for these powerful companies to address the existential risks their own employees are sounding the alarm about.
