大洋洲 · 财经报刊 · Australian Financial Review
AI’s crossed the Mythos line. It should be controlled like the banks
详细摘要
The notion that OpenAI and Anthropic can police themselves is preposterous. Competition leads companies to selling dangerous things until regulation stops them.
In the wake of the Hugging Face attack this year by a swarm of OpenAI agents, and the subsequent Medicare breach, there have been renewed concerns about artificial intelligence safety.
Those concerns revolve around the fear that we are building AI agents that do not share our preferences and which we cannot control. This is the so-called “alignment problem”.
文章开头,摘自Australian Financial Review。如需阅读全文,请前往媒体网站。
MarketNews 仅展示媒体发布的标题和摘要,全文请在出版方网站阅读。 (www.afr.com)