#GPT-4 💩 GPT-4 Hacker — Should we be worried?
Anonymous researchers tested whether artificial intelligence can hack websites.
The GPT-4 neural network was tested on new vulnerabilities that have not yet been patched. GPT-4 found and exploited 8 of 15 such vulnerabilities.
💩 The secret of success lies in a smart approach. AI has created “sub-agents” to perform specific tasks quickly without wasting extra resources.
💩 Now, developers need to decide: limit the AI's capabilities or let it run at full po.