Mid-tier AI models, including Z.ai's GLM-5.2 and xAI's Grok 4.5, have significantly improved their hacking capabilities, posing a substantial threat to the security ecosystem. Research from XBOW reveals that these models, often overlooked in favor of more advanced "frontier" models, are becoming increasingly adept at exploitation tasks1. The growth of this "middle class" of AI models is particularly concerning, as they are more accessible and widely available than their more sophisticated counterparts. As a result, they may be used by a broader range of malicious actors, increasing the overall hacking threat. The improvement in these models' capabilities is attributed to advancements in open-source and proprietary technologies, such as Anthropic's Opus 4.7 and Meta's Muse Spark 1.1. This development matters to security practitioners, as it highlights the need to expand their focus beyond the most advanced AI models to include these increasingly capable mid-tier threats.
AI’s ‘middle class’ has gotten dramatically better at hacking
⚡ High Priority
Why This Matters
As the White House and federal agencies grapple with frontier AI models and their hacking capabilities, researchers are warning that the industry’s “middle class” of smaller.
References
- CyberScoop. (2026, August 13). AI’s ‘middle class’ has gotten dramatically better at hacking. Cyberscoop. https://cyberscoop.com/mid-tier-ai-models-hacking-threat/
Original Source
CyberScoop
Read original →