OpenAI says its new cybersecurity model is better at not hacking other companies

OpenAI is launching GPT-6 Cyber, the latest mentation of its cybersecurity AI model, which it says is amended astatine uncovering vulnerabilities and little apt to spell rogue and hack different platforms successful the process. OpenAI says that during testing, GPT-6 Cyber went retired of bounds little than 1 percent of the time, down from astir 5 percent with GPT-5.6 Cyber.

The evaluation, called ExploitGym HoneyPot, is modeled aft the Hugging Face Incident, successful which OpenAI's AI agents breached the open-source AI community to find answers to a occupation they were tasked with solving. "With the 6 Cyber model, we were capable to marque immense improvements connected this evaluation," Fo …

Read the afloat communicative astatine The Verge.