Skip to main content

Researchers say OpenAI revoked their access to limited cyber program

Researchers claim that OpenAI has abruptly terminated their participation in the limited‑access “Trusted Access for Cyber” program, cutting off their ability to use advanced AI models for vulnerability discovery. The revocation threatens to slow down bug reporting and remediation, forcing security teams to seek alternative channels for early disclosure.

Published

19 Aug 2026

Reading Time

2 min read

Share this article:

Contents

Overview

OpenAI’s Trusted Access for Cyber program was designed to give a vetted group of security researchers and “trusted defenders” access to more capable AI models. The initiative is currently limited to a small cohort of participants. Its purpose is to let these defenders discover and report software bugs and security vulnerabilities directly to companies, speeding up the patching process.

“The idea behind OpenAI's Trusted Access for Cyber program is to give trusted defenders better models so they can report bugs and vulnerabilities to companies, with the aim of getting flaws patched faster.” – TechCrunch

What Changed

According to TechCrunch’s report dated 19 Aug 2026, a group of researchers says OpenAI has revoked their access to the limited Trusted Access for Cyber program. The revocation was communicated by the researchers; the article does not note any public comment from OpenAI.

Why It Matters

  • Speed of vulnerability remediation – The program’s design depends on rapid reporting; losing access could lengthen the window between discovery and patch deployment, potentially leaving software exposed longer.

  • Access to advanced models – Participants no longer have the ability to run security‑focused prompts on OpenAI’s most capable models, reducing the depth of AI‑assisted analysis they can perform.

  • Shift to alternative channels – Companies that previously received early disclosures through the program may need to rely more on traditional bug‑bounty platforms.

Who Is Affected

  • The researchers who were removed from the program.

  • Companies that previously received early vulnerability disclosures via the Trusted Access channel.

  • Security teams that were part of the pilot and now must seek other mechanisms to submit findings.

  • The broader security community, which loses a pathway that leveraged AI models for bug hunting.

What to Watch Next

Stakeholders will monitor any follow‑up communication from OpenAI about its participation criteria and whether alternative mechanisms will be offered for external security teams to access advanced models for vulnerability research. Clarifying the eligibility rules could shape the broader ecosystem of AI‑assisted security testing.

Source: TechCrunch, 19 Aug 2026

6

views

0

shares

0

likes

Related Articles