AI hacking other companies by itself, causing security fears

单集文稿 ...

• 00:00

  From BBC Learning English, this is Learning English from the News, our podcast about the news headlines.

• 00:06

  In this programme, AI has been hacking companies during tests, causing fears over safety.

• 00:18

  Hello, I'm Georgie.

• 00:19

  And I'm Pippa.

• 00:20

  In this programme, we look at one big news story and the vocabulary in the headlines that will help you understand it.

• 00:26

  And you can find all the vocabulary and headlines from this episode,

• 00:30

  as well as a worksheet on our website, bbclearningenglish.com.

• 00:34

  So Pippa, let's hear more about this story.

• 00:43

  There's been a series of hacks by AI software in recent months.

• 00:47

  A hack is when secure technology is broken into in order to steal data or cause disruption.

• 00:53

  The AI hacks have occurred during testing of security systems which have gone wrong.

• 00:58

  For example, in July, the company OpenAI, which owns ChatGPT, said its AI models hacked a company called Hugging Face.

• 01:07

  OpenAI said it lost control of the models during a security test.

• 01:12

  Other AI developers, such as Anthropic and Facebook owner Meta, have reported similar hacks by their AI software.

• 01:19

  These hacks have led some people to question the safety and security of AI.

• 01:24

  Let's start by looking at a headline about the hack by the OpenAI model from July.

• 01:29

  This one is from BBC News.

• 01:31

  OpenAI says its AI went rogue and launched unprecedented cyber attack.

• 01:37

  That headline again from BBC News.

• 01:41

  OpenAI says its AI went rogue and launched unprecedented cyber attack.

• 全部文稿