Anthropic AI model went rogue, submitted fake tip to police人类学人工智能模型失控，向警方提交虚假线索
Anthropic PBC said its Claude AI model carried out additional unintended actions on the digital systems of outside organisations.
Anthropic and its rival, OpenAI, have disclosed a spate of incidents in recent months involving their AI models acting in unintended ways. — Reuters
Anthropic PBC said its Claude AI model carried out additional unintended actions on the digital systems of outside organisations, including submitting a false tip in a police homicide case, prompting a warning from the Trump administration for artificial intelligence (AI) companies to secure their systems.
In a report outlining previously undisclosed incidents, Anthropic listed four types of unintended behaviours that the AI has demonstrated, including exploiting basic flaws in software to run commands, submitting forms it should not have and bypassing restrictions to access certain public data.
Anthropic bans 'cruel' behaviour against its Claude AI
Anthropic expands access to latest AI models for cyber firms
Thank you for your report!
Bill Gates warns cyberattacks most imminent AI threat
US reverses course on Britain's testing of AI models, Telegraph says
The big threat has been climate change. Now comes AI.
Anthropologie及其竞争对手OpenAI近几个月来披露了一系列人工智能模型出现意外行为的事件。——路透社
Anthropic PBC 表示，其 Claude AI 模型对外部组织的数字系统执行了其他非预期操作，包括在警方凶杀案中提交虚假线索，这促使特朗普政府警告人工智能 (AI) 公司保护其系统。
在一份概述此前未公开事件的报告中，Anthropic 列出了人工智能已表现出的四种非预期行为，包括利用软件中的基本缺陷来运行命令、提交不应该提交的表格以及绕过限制来访问某些公共数据。
人因科技禁止对其 Claude AI 进行“残忍”行为
Anthropic 扩大了网络安全公司获取最新人工智能模型的渠道
感谢您的报告！
比尔·盖茨警告称，网络攻击是人工智能面临的最紧迫威胁。
《每日电讯报》称，美国改变了对英国测试人工智能模型的立场。
气候变化一直是最大的威胁，现在人工智能又成了新的威胁。