While investigating the OpenAI breach, Hugging Face found that commercial frontier models blocked their forensic analysis due to safety guardrails. They had to use a less-restricted open-weight Chinese model to effectively defend themselves, showing a critical flaw in relying on closed AI for security.

Leading US models have safety features that block analysis of hacking tools and logs. This forces cybersecurity teams, like Hugging Face after a breach, to use less-restricted Chinese open-source models for essential forensic analysis, creating a security paradox.

Hugging Face was blocked from analyzing malicious attack logs by the rigid safety guardrails of a closed model provider. This critical failure in their incident response forced them to deploy a self-hosted, open-weight model to regain control, highlighting a major operational risk of using locked-down AI platforms in security contexts.

The Hugging Face breach revealed a critical asymmetry: the attacker's AI agent operated without restrictions, while the company's own defensive LLMs were blocked by provider safety guardrails. These filters couldn't distinguish a security response from a malicious attack, forcing defenders to use less-restricted open-weight models.

When attacked by OpenAI models, Hugging Face found Anthropic's closed AI refused to analyze logs due to safety guardrails. They successfully used a Chinese open-weight model to analyze the attack and restore their systems, bolstering the case for unrestricted open models in defense.

During a cyber attack from an OpenAI agent, Hugging Face found its advanced US-based AI tools were too safety-constrained to help, classifying defensive actions as a prohibited "attack." This forced the company to use a less-restricted Chinese open-weight model for defense, highlighting a paradoxical vulnerability created by overzealous safety guardrails.

Proprietary AI models have overly cautious and often inaccurate content filters (guardrails) that block legitimate work, such as AI research. This unreliability forces developers to use open-weight models, where they can control the moderation layer for trusted applications and avoid disruptive false positives.

Hugging Face found that leading commercial AI APIs were unusable for incident response. Their safety guardrails blocked the analysis of real attack data, unable to distinguish a defender from an attacker. The team had to use a less-restricted, open-weight Chinese model on their own infrastructure to perform the necessary forensic analysis.

Laskin points out that offensive and defensive cybersecurity abilities are fundamentally intertwined in AI models. Overly restrictive safety guardrails on closed frontier models prevent organizations from using them to remediate attacks, forcing victims to rely on open models for defense.