OpenAI 'ethically hacked' with help of Anthropic's Claude chatbot

11 pointsposted 5 hours ago
by sbulaev
(theguardian.com)

2 Comments

kittikitti

5 hours ago

Here's an important detail that the title of the article obfuscates, "the researchers said they were largely using OpenAI’s own cutting edge GPT-5.6 Sol model to carry out the hack".