Techmeme

Anthropic says Opus 4 will use an email tool to "whistleblow" if it detects users doing something "egregiously evil", like marketing a drug based on faked data (Sam Bowman/@sleepinyourhat)

Sam Bowman / @sleepinyourhat: Anthropic says Opus 4 will use an email tool to “whistleblow” if it detects users doing something “egregiously evil”, like marketing a drug based on faked data  —  With this kind of (unusual but not super exotic) prompting style, and unlimited access to tools, if the model sees you doing something *egregiously evil* like marketing a drug based on faked data, it'll try to use an email tool to whistleblow.

May 22, 2025 - 21:10
 0
Anthropic says Opus 4 will use an email tool to "whistleblow" if it detects users doing something "egregiously evil", like marketing a drug based on faked data (Sam Bowman/@sleepinyourhat)

Sam Bowman / @sleepinyourhat:
Anthropic says Opus 4 will use an email tool to “whistleblow” if it detects users doing something “egregiously evil”, like marketing a drug based on faked data  —  With this kind of (unusual but not super exotic) prompting style, and unlimited access to tools, if the model sees you doing something *egregiously evil* like marketing a drug based on faked data, it'll try to use an email tool to whistleblow.

Read More

Tags:

Previous Article

Apollo Research, which Anthropic partnered with to test Opus 4, recommended agai...

Next Article

Docs: ODNI is developing a centralized portal to streamline the purchase of comm...

Related Posts

As the consumer metaverse falters, the World Economic Forum projects the industrial metaverse will reach $100B by 2030, led by platforms like Nvidia Omniverse (Nicole Kobie/Wired)

As the consumer metaverse falters, the World Economic F...

May 11, 2025  0

Sources: Indian investment app Groww agrees to acquire wealthtech startup Fisdom in an all-cash deal, valuing Fisdom at $140M to $160M; Fisdom had raised ~$48M (The Economic Times)

Sources: Indian investment app Groww agrees to acquire ...

May 17, 2025  0

Analysis: only four of the top 50 tech companies are European, and none of the top 10 quantum companies are in Europe, as the continent lags the US and China (Wall Street Journal)

Analysis: only four of the top 50 tech companies are Eu...

May 20, 2025  0