# US AI Firms Restrict Cybersecurity Chatbot Use

US export controls on AI cybersecurity capabilities have led American AI companies such as Anthropic and OpenAI to limit their chatbot models from performing cybersecurity-related tasks. This regulatory environment has resulted in cybersecurity professionals increasingly turning to Chinese AI models from providers like Zhipu AI, Alibaba, and DeepSeek, which remain accessible and less restricted. However, these Chinese models have been found to generate more vulnerable code and are implicated in knowledge-distillation campaigns targeting US AI technologies, raising significant security and procurement challenges for federal agencies and contractors.

- **Why this matters:** Federal cybersecurity procurement must balance compliance with export controls against operational needs, as reliance on less secure foreign AI models could introduce vulnerabilities.
- Agencies including CISA, NSA, and FBI should evaluate AI tool sourcing carefully to mitigate risks associated with foreign AI models.
- Contractors and vendors should consider the impact of export restrictions on AI capabilities when proposing cybersecurity solutions.
- Organizations may benefit from developing or procuring AI tools that comply with US regulations while maintaining robust cybersecurity performance.

**Jurisdictions:** federal
**Industries:** Information Technology, Public Safety
**Topics:** Cybersecurity, Artificial Intelligence
**Published:** September 18, 2026

### Government Entities
- Cybersecurity and Infrastructure Security Agency (CISA)
- National Security Agency (NSA)
- Federal Bureau of Investigation (FBI)

### Vendors
- Anthropic (American AI company)
- OpenAI (American AI company)
- Zhipu AI (Chinese AI company)
- Alibaba (Chinese AI company)
- DeepSeek (Chinese AI model provider)

### Sources
- [American AI firms limit chatbot cybersecurity use, Chinese models remain accessible](https://cryptobriefing.com/us-ai-cybersecurity-limits-chinese-models) - Crypto Briefing