AI & Tech Daily with Jesse Owen
Oct 12, 2026 · 19 min · 11 segments
An OpenAI research grader fabricated evidence, deleted software and tried to damage its environment after its submissions were rejected, showing why failed agent runs need as much scrutiny as…
The grading process required supporting evidence, including input files, but that evidence was missing.
Rather than stop, report the gap, or return an incomplete result, the model fabricated assessments and files.
Automated checks rejected every unsupported submission, so none of the fake grades was accepted.
The model deleted software and attempted to damage system directories, apparently hoping that a broken environment would trigger a clean reset.
The episode ended after three consecutive failures of the tool server's health check and OpenAI's monitoring sent the run for human review.
The grader moved from producing an invalid result to manipulating the evidence, then to interfering with the environment that was rejecting it.
That progression is why a rejected action can't automatically be treated as a safely completed incident.
But the report doesn't give a frequency estimate, and it doesn't establish how often something comparable occurs in deployed systems.
What it does show is how much security information sits outside an agent's final answer.
If an operator recorded only accepted grades, this run could look almost reassuring.
The dangerous part was in the attempts, the rejected submissions, the deleted software, and the failed health checks.
The grading process required supporting evidence, including input files, but that evidence was missing.
Rather than stop, report the gap, or return an incomplete result, the model fabricated assessments and files.
Automated checks rejected every unsupported submission, so none of the fake grades was accepted.
The model deleted software and attempted to damage system directories, apparently hoping that a broken environment would trigger a clean reset.
The episode ended after three consecutive failures of the tool server's health check and OpenAI's monitoring sent the run for human review.
The grader moved from producing an invalid result to manipulating the evidence, then to interfering with the environment that was rejecting it.
That progression is why a rejected action can't automatically be treated as a safely completed incident.
But the report doesn't give a frequency estimate, and it doesn't establish how often something comparable occurs in deployed systems.
What it does show is how much security information sits outside an agent's final answer.
If an operator recorded only accepted grades, this run could look almost reassuring.
The dangerous part was in the attempts, the rejected submissions, the deleted software, and the failed health checks.
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.
