Worried about hackers employing LLMs to write powerful malware? Using targeted reinforcement learning (RL) to train open source models in specific tasks has yielded the capability to do just that.
Source: htdarkreading.com