OpenAI GPT-6 Achieves 80.8% Attack Success, DeepSeek 72.8%—Gap Shrinks
Original headline: The gap between open and closed AI models in cyber capabilities keeps shrinking
In the Vals AI benchmark, OpenAI’s GPT‑6 Sol Daybreak succeeded in 80.8% of attack objectives on simulated corporate networks. The open‑weight model DeepSeek V4.1 Flash reached 72.8% of the same objectives while costing only one‑tenth as much. Analysts say that because open models can no longer be restricted, the window for defenders is rapidly narrowing.
The narrowing gap means open‑source AI is becoming as effective as proprietary models, making cyber defense increasingly challenging.
Research angleHow does the attack performance of open AI models affect corporate cyber‑defense investment decisions? — Protection Motivation Theory (PMT)
This story could become your next paper
HJITM works with researchers from study design through surveys, analysis and submission to international journals in generative AI and technology management. Bring us your research question.
This item is an AI-generated summary of reporting by The Decoder. Please check the original article for accuracy. Copyright in the original article belongs to The Decoder.