Published

OpenAI partners with Hacktron to find system vulnerabilities, focusing on third-party service connections and expanding Daybreak Red access

According to Unite.AI’s write-up of OpenAI’s announcement, OpenAI said on 6 November it is partnering with Hacktron to find vulnerabilities in its systems, concentrating on connections with third-party services and expanding the firm’s access to frontier models through Daybreak Red. The backdrop includes a cross-product exploit chain Hacktron disclosed from July.

Abstract pale-cyan swirling line illustration on a dark blue background
Cover art from Hacktron’s “Hacking OpenAI” post: a pale-cyan line-work vortex on a dark blue background.

Unite.AI’s account of the OpenAI announcement says OpenAI engineers and Hacktron researchers will jointly pick high-priority areas to test, examining weaknesses that appear when AI agents link applications, identity systems and infrastructure in new ways, with lessons used to harden defenses. The partnership also expands Hacktron’s model access through Daybreak Red: Red is for authorized vulnerability research, exploit validation and security testing, while Blue is the starting point for most defenders. Access is limited to approved individuals and organizations, with identity verification, account security, monitoring, approved-use restrictions and legal attestations.

Unite.AI says OpenAI came to know Hacktron through its Security Bug Bounty program and was impressed by the impact of vulnerabilities the firm discovered and disclosed. Hacktron describes building AI agents for code review, penetration testing and red teaming. As a concrete prior case, Hacktron’s blog says that on 25 July 2026 it chained remote code execution in the Discourse-hosted community forum’s image-processing path with an OpenAI single sign-on flaw to reach multiple employees’ ChatGPT and Codex accounts and, in theory, connected integrations such as GitHub; to prove access without reading sensitive internal code, the team used an employee’s Codex account to open a harmless pull request in OpenAI’s internal monorepo, then stopped testing and reported via Bugcrowd to OpenAI and to Discourse. Hacktron’s timeline says OpenAI confirmed a fix about 14 hours after the initial report and Discourse had a fix ready two days after receiving its report; OpenAI paid a $6,500 bounty on 1 September 2026 while clarifying that the award recognized the OpenAI-side finding rather than testing against the explicitly out-of-scope Discourse-hosted forum. Vulnerability details and the bounty amount are Hacktron’s / secondary claims.

Sources

  1. OpenAI Partners With Hacktron to Test Third-Party Service Connections — Unite.AI
  2. Hacking OpenAI — Hacktron
Open this story in Shift →