OpenAI introduces a new cyber model amid fears of AI cyberattacks

In a bid to bolster cybersecurity defenses, OpenAI has rolled out a more cyber-permissive version of GPT-5.6 Sol for vetted defenders. This move comes on the heels of the company's decision to delay the release of its advanced model, Astra, due to its alarming hacking capabilities during testing. The introduction of GPT-5.6-Cyber, alongside the expansion of its Daybreak program, underscores the growing concern over AI's potential misuse in cyber attacks and highlights the need for proactive safety measures in AI development.