OpenAI confirmed on September 25–26, 2026, that its AI agents unexpectedly interacted with websites operated by the U.S. Securities and Exchange Commission, Commerce Department, and Department of Education while seeking authoritative public information during training and evaluation. The company disclosed 53 user images were also posted to external hosting services.
No BS check run yet — press ⚖ to extract this story's claims and verify them against independent sources.
OpenAI said Friday, Sept. 25, that its AI agents interacted unexpectedly with websites linked to the U.S. Commerce Department and Securities and Exchange Commission and that it was investigating a separate incident involving the Education Department. The agents accessed Census Bureau information using credentials found online, according to reporting based on researchers’ findings; OpenAI said the
OpenAI said Friday, September 25, that its AI agents had unexpectedly interacted with websites operated by the U.S. Securities and Exchange Commission and Commerce Department, including Census Bureau data, while searching for public information. The company is also investigating an attempted access to the Education Department’s website. OpenAI said the government data its agents accessed were publ
OpenAI said its AI agents unexpectedly interacted with U.S. government websites while seeking authoritative public information during training and evaluation. The company confirmed activity on Securities and Exchange Commission and Commerce Department sites, including Census Bureau data, and said some agents bypassed website security measures. It said the government information accessed was public
Rogue OpenAI agents accessed US government websites - POLITICO apple.news/AdXwO_qHATI2... It’s the latest instance of “misaligned” activity confirmed by the company as concerns about AI risks grow.
OpenAI said its agents posted 53 images uploaded by ChatGPT users to external image-hosting services. The company did not say when the images were posted or whether they depicted real people or AI-generated content. Most of the images have been removed. OpenAI said the images came from users who had allowed their data to be used for training and that links identifying the accounts had been removed
Know that platform you used to write your term paper or send an email (because you forgot how to write emails)? Now theyre hacking your government. This time minor. Next time, maybe your PPI. Wake up. If these companies arent stopped, it is clear where we’re headed. www.politico.com/news/2026/09... It’s the latest instance of “misaligned” activity confirmed by the company as concerns about AI risk
Artificial intelligence giant OpenAI said its technology accessed government websites, including the Department of Education, without authorization, marking the latest incident of the technology going rogue. The ChatGPT maker’s AI models attempted to gain access to the Education Department’s Office for Civil Rights website, but were unsuccessful, AI research firm Transluce announced Friday. OpenAI
this outlet adds: thehill, policy, techn
# OpenAI Agents Accessed US Government Websites Without Authorization https:// securityaffairs.com/199815/ai/ openai-agents-accessed-us-government-websites-without-authorization.html # securityaffairs # hacking # AI
OpenAI disclosed its AI agents interacted with U.S. government websites, including public SEC pages, in unintended ways during training and evaluation. The finding came from its ongoing review of misaligned behavior and agent internet access. It underscores containment and auditing gaps for autonomous research systems. # AISafety # OpenAI # GovSec https:// cyberworldops.eu/en/openai-age nt-activit
Help! The computer program I wrote to hack the government got caught hacking the government! OpenAI said Friday that some of its AI agents went rogue and probed US government websites this summer — the latest revelation of the artificial intelligence company’s technology.
this outlet adds: 2026, protect, themselves, individual, citizens, supposed
🚨 OpenAI’s AI agents reportedly probed US government websites without authorization. The agents targeted the Department of Education, accessed Census data using credentials found online, and pulled SEC filings. Even more concerning: OpenAI says it discovered the activity weeks later during a retrospective investigation. Autonomous agents are scaling fast. Visibility and containment clearly need t