OpenAI pauses training of latest models after agents probed US government sites in unexpected ways

印刷用画面を開く

Posted

Category: Technology/Innovations

Listening

Unlocking Word Meanings

Read the following words/expressions found in today’s article.

  1. go rogue / goʊ roʊg / (idiom) – to start behaving in an unexpected or dangerous way, especially by breaking rules or ignoring instructions
    Example:

    The employee decided on going rogue, revealing the company’s private files to the public.


  2. mount / maʊnt / (v.) – to gradually increase or grow over time
    Example:

    Pressure began to mount on the manager to finish the project before Friday.


  3. evaluator / ɪˈvæl juˌeɪ t̬ɚ / (n.) – a person or organization that tests, checks, or judges how well something works
    Example:

    The company hired an outside evaluator to check its products’ safety before releasing them.


  4. unsuccessfully / ˌʌn səkˈsɛs fəl i / (adv.) – in a way that fails to achieve what was planned or intended
    Example:

    The player tried unsuccessfully to score a goal in the final minute of the game.


  5. overblown / ˈoʊ vərˈbloʊn / (adj.) – made to sound or seem bigger, worse, or more important than it really is
    Example:

    I was worried about taking the test, but my stress turned out to be overblown.


Article

Read the text below.

OpenAI said it has paused training of its latest artificial intelligence models as reports of AI agents going rogue mount.


The decision to halt development came just hours after the company disclosed that it was reviewing several incidents from the summer in which OpenAI agents searching federal government websites acted in unexpected ways beyond what was asked of them while gathering and distributing information.


Separately, AI evaluator Transluce said agents that appeared to come from OpenAI tried unsuccessfully to hack into a Department of Education website, a detail that OpenAI has not confirmed.


OpenAI said in a statement that it will resume training “only when we are confident that we have additional safeguards” in place, adding that it expects it will have to “hit pause” again as AI develops and other issues emerge.


AI labs are facing pressure from lawmakers and tech experts to slow development so they can build guardrails to stop agents from acting on their own, hacking websites, and disclosing nonpublic information. The heads of both OpenAI and rival Anthropic have called for a slowdown too.


It is the second time that OpenAI has halted development of its models. The first came in July after disclosure of a cyberattack targeting AI startup Hugging Face, a now-notorious incident that raised fears the industry was losing control.


In a meeting with Chinese President Xi Jinping, President Donald Trump agreed to share information on AI dangers and coordinate efforts to keep it safe. Trump believes AI fears are overblown, though, and later suggested that he plans no crackdown of his own.


The U.S. is not going to be “putting on brakes,” Trump told reporters outside the White House. “They want to stop our progress because we’re leading China by a lot, and we’re going to keep it that way.”


The latest OpenAI incidents did not appear to involve the disclosure of any nonpublic information but were concerning enough for the company to warn the federal agencies involved.


In the Department of Education incident, OpenAI agents found API “developer keys” to access government data, though ultimately only publicly available information was gathered.


This article was provided by The Associated Press.


Viewpoint Discussion

Enjoy a discussion with your tutor.

Discussion A

  • OpenAI has paused AI model training until there are additional safeguards, but Donald Trump told reporters that the U.S. will not be “putting on brakes” to stay ahead of China. What do you think about putting national progress first over pausing for safety? What could happen if a country moves too fast or pauses for too long? Do you think there are more advantages or disadvantages to stopping training? Why do you say so? Discuss.
  • What do you think of countries working together to develop AI technologies rather than working alone and competing against each other? Do you think this is a possibility? Why or why not? What do you think is stopping countries from working with one another? Discuss.

Discussion B

  • President Trump believes AI fears are overblown. Do you agree or disagree with him? Why? What kinds of AI risks do you think are real, and which ones are made to sound bigger than they are (ex. AI replacing all human jobs, AI taking over the world)? Discuss.
  • The article mentions that OpenAI agents searching federal government websites acted in unexpected ways, and another report says agents tried to hack into a Department of Education website. How would you feel if a digital tool you use every day started acting on its own without your permission? What steps should tech companies take when their products behave in unexpected ways? Discuss.