# Putting frontier cyber models in more trusted hands

The expanded program includes Accenture, IBM, CrowdStrike, Palo Alto Networks, Cisco, Sophos, Cloudflare, and more as partners.

- Partners can access Daybreak Blue for defensive workflows and Daybreak Red for red teaming and penetration testing through Daybreak Access.
- Other participants include Capgemini, Cognizant, EY, KPMG, PwC, NCC Group, SpecterOps, Unit 42, Fortinet, and Akamai.
- Access to the underlying models remains with the approved partner and is not transferred directly to the customer.
- Safeguards can include identity verification, defined testing scopes, logging, monitoring, and human oversight.

## Why it matters

Organizations can now get frontier OpenAI cyber capabilities through their existing security providers and consultants instead of building specialized cyber AI programs.

## Sources

- [OpenAI: Putting frontier cyber models in more trusted hands](https://openai.com/index/putting-frontier-cyber-models-in-more-trusted-hands)

---

Summarized by dstilled on 2026-08-05. https://dstilled.ai/story/354fc912-6cdd-4007-b9aa-4c1f6a3b98e2