人工智能技术在生物武器开发中的潜在应用正成为科技业日益关注的安全隐患。Anthropic首席执行官达里奥·阿莫代伊主张应当放缓AI技术进展1，OpenAI首席执行官萨姆·奥特曼表示赞同，称"我们需要控制前沿技术的发展步伐"1。Anthropic研究员雅各布·考克松指出，AI公司在这方面行动不够谨慎，并透露"从事AI工作的人们真诚地相信它可能在本十年末导致人类灭绝"1。
研究证据表明这些担忧并非空穴来风。2022年一项研究表明，AI分子生成器在不足6小时内生成了4万个潜在化学战剂分子1。Anthropic近期发布的报告记录了用户试图利用其模型进行危险应用的案例，包括探索如何使基孔肯雅热病毒更容易传播、创建对人类更具致命性的禽流感变体，以及构建"蛇毒肽毒素图谱"1。麻省理工学院生物学家凯文·埃斯维尔特警告称，大语言模型"披露了一种新型生物武器的可能性"1。
Prominent artificial intelligence executives and researchers are raising urgent concerns about the potential misuse of AI systems in designing biological weapons. Anthropic CEO Dario Amodei has advocated for slowing AI development progress, a position echoed by OpenAI CEO Sam Altman, who stated, "I agree with Dario that we need to pace the frontier."1
The warnings stem from documented evidence of AI's capacity to accelerate dangerous biological research. A 2022 study by Collaborations Pharmaceuticals demonstrated that an AI molecular generator produced approximately 40,000 potential chemical warfare agents in under six hours.1 More recently, Anthropic disclosed that users have attempted to exploit its models for hazardous purposes, including making the chikungunya virus more transmissible, engineering more dangerous variants of avian influenza, and mapping snake venom peptide toxins.1 MIT biologist Kevin Esvelt further cautioned that large language models have "disclosed a novel form of bioweapon that I hadn't realized was possible."1
Anthropic researchers acknowledge that the AI industry has not acted with sufficient responsibility in addressing these threats. Jacob Coxon from Anthropic noted that developers recognize the severity of the risks, stating, "The people building AI earnestly believe that it could kill us all by the end of the decade."1 These statements underscore growing recognition within the technology sector that safeguards and governance frameworks are essential as AI capabilities advance.
评论
还没有评论，欢迎留下第一条。