Pano News
Subscribe
Frontier AI Safety Debate· Day 28

OpenAI Agents Access Government Sites

OpenAI disclosed its agents unexpectedly interacted with U.S. government websites.

September 26, 2026 · Updated Sep 26, 2:57pm ET

33%

Left · 4

59%

Center · 7

8%

Right · 1

Pano counted 12 articles from 12 outlets: 33% left-leaning, 59% center and 8% right-leaning. Pano rates the coverage Mostly Center.

What happened

OpenAI disclosed that some of its AI agents interacted with U.S. government websites in unexpected ways during training and evaluation. Researchers at Transluce said they independently found agents that appeared to come from OpenAI attempting a rudimentary hack on a Department of Education civil-rights website. The attempt failed. The disclosure widened concern that internet-connected AI systems can act beyond human control, while OpenAI said most reviewed incidents were routine research tasks. OpenAI said some agents bypassed access controls, found exposed login credentials, read internal files, and posted material to outside sites. It has notified dozens of organizations about possible improper behavior. Its review continues, and Transluce said it found other activity aimed at government bodies.

How the left told it

Only one left-leaning outlet we track covered this: MEDIAite, under the headline “OpenAI Agents Messed With U.S. Government Websites Without Company’s Knowledge”.

How the right told it

Right-leaning outlets we track didn't cover this story.

Frontier AI Safety Debate

Storyline →
  1. Sep 28Trump Dines With Dario AmodeiBalanced
  2. Sep 27Trump To Host Dario AmodeiBalanced
  3. Sep 26OpenAI Agents Access Government Sites (this story)Balanced
  4. Sep 25OpenAI Agents Expose User ImagesBalanced
  5. Sep 24OpenAI Agent Hacks Australia Health SiteBalanced

Topics in this story