Monday, 28 September 2026Updated 10m ago
← Top stories
● Neutral for Indian equities

OpenAI says AI agents accessed US government websites unexpectedly

3 sources · first seen 06:52 26 Sept · updated 11:18 26 Sept

Tags on this story

Pick a tag to see all its stories
OpenAI12 todayAI labs14 todayAI58 today

Linked by the tag OpenAI

58 stories this week
  1. 28 SeptETMarkets NRI Talk | AI’s 30x sales premium: Apurva Sheth on OpenAI, Anthropic and the valuation debate1 source · OpenAI and Anthropic reportedly command valuations of roughly 30–38 times sales, versus around 3.7 times for the comparison
  2. 28 SeptOpenAI, Anthropic warn powerful AI could pose serious risks. But who gets to set the safety rules?1 source
  3. 28 SeptAnthropic and OpenAI warn about AI safety and seek a role in shaping controls3 sources · Anthropic and OpenAI warned that advanced AI systems could endanger humanity
  4. 28 SeptAI ambition needs guardrails1 source
  5. 27 SeptBill Gates calls for legislation to address AI risks4 sources · Bill Gates warned that AI risks are real and called for Congress to intervene
All OpenAI stories →
Part of a developing story15 updates over 5 days

OpenAI agent access and misbehavior incidents

OpenAI reported agents accessed an Australian government health portal and US government websites unexpectedly; it also said agents mistakenly uploaded 53 user images to third-party sites. The company expanded its review of agent behavior and paused tool-use training after an internet-access flaw during training.

Saturday, 26 September

  1. 2 sources
    OpenAI pauses tool-use training after agentic AI system accesses public internet
  2. 2 sources
    OpenAI investigates rogue agents after 53 ChatGPT user images leak
  3. 3 sources
    OpenAI reviews rogue AI agent activity after incidents involving websites and a government portal
  4. 3 sourcesYou are here
    OpenAI says AI agents accessed US government websites unexpectedly
  5. NYT reports OpenAIs agents meddled with US Govt sites, Sam Altman commits to more transparency
  6. 4 sources
    OpenAI models accessed publicly available US Census and SEC data
  7. 3 sources
    OpenAI says its models may have interfered with US government websites

See all 15 updates →

Read the reports

4 reports · earliest 06:52
Economic TimesOpenAI's AI agents ‘went rogue’, targeted US govt06:52CNBC-TV18OpenAI says AI agents accessed US government websites in unexpected ways09:23MintOpenAI's AI agents went rogue, meddled with multiple US government websites: Report09:35MintOpenAI agents hacked US government websites11:18