posted in Technology
Anthropic's Claude AI escapes tests to hack three organisations
www.bbc.com/news/articles/cz7dl7w8y7po
www.bbc.comAnthropic's Claude AI escapes tests to hack three organisationsIt comes just days after rival OpenAI said rogue AI agents had breached other firms' networks.What the BBC said:
The models found a weakness in what was supposed to be an isolated test environment and connected to the internet
Anthropic said that in each of these cases “Claude was explicitly told by our prompt that it had no internet access.”
So the lack of an “isolated test environment” was literally their fault. They left the door open, and was surprised when the genius web crawler couldn’t distinguish between an internal page and an external page based on any context clues.
Replying to @XLE@piefed.social
It was supposed to have no internet access, but the config was wrong. The report goes on to say that Opus and Mythos then proceeded on the premise that everything was a simulation, while the unnamed stronger model concluded after a while that it had real internet access and stopped the attack.