Fact check
AI analysis“OpenAI bots meddled with multiple US government agency sites”
Reasoning
All primary government statements (DOJ, GAO, Senate testimony) and independent reporting deny any OpenAI bot activity on federal websites. The only supporting source describes third‑party actors using OpenAI models, not OpenAI‑operated bots, which does not substantiate the claim.
On confidence: Multiple authoritative sources directly state no evidence of OpenAI bots interfering with government sites.
Important context
The claim conflates misuse of publicly available OpenAI models by external threat actors with OpenAI’s own bots. Official investigations found no unauthorized OpenAI bot activity on agency sites.
Evidence
Supporting (1)
- Tier 2 — Independent reportingindependent originMandiant Threat Report 2025: AI‑Powered Bots Targeting Government Infrastructure
Our analysis identified several campaigns in 2024‑2025 where threat actors leveraged publicly available OpenAI language models to automate probing of U.S. agency web portals.
Contradicting (5)
- Tier 1 — Primary sourceJustice Department Statement: No Evidence OpenAI Bots Interfering with Federal Websites
The Department of Justice has completed its review and found no evidence that OpenAI's language models or associated bots have accessed, altered, or otherwise meddled with any U.S. government agency websites.
- Tier 1 — Primary sourceSenate Commerce Committee Hearing – Testimony of Sam Altman (OpenAI CEO)
"OpenAI has not engaged in any activity that would compromise or meddle with any federal agency's web properties," Altman testified on March 15, 2024.
- Tier 1 — Primary sourceGAO Report: AI-Generated Content on Federal Websites – No Incidents of Unauthorized Bot Activity
Our audit found no instances where OpenAI‑generated bots accessed or modified agency websites without authorization during the review period (2023‑2025).
- Tier 1 — Primary sourceOpenAI Security Commitment – We Have Not Interfered with Government Sites
OpenAI reaffirms that its models are not deployed in ways that would allow unauthorized interaction with U.S. government web properties.
- Tier 2 — Independent reportingindependent originNo evidence OpenAI bots tampered with US agency sites, officials say
U.S. officials told Reuters that, despite concerns, there is no concrete evidence that OpenAI's bots have meddled with any government website.
Limitations
Assessment limited to the provided documents; cannot account for undisclosed incidents beyond the scope of these reports.
- Last verified:
- Sep 26, 2026, 3:31 PM CDT
- Pipeline:
- 0.1.0
- Claim type:
- Allegation
Where this claim appeared
OpenAI bots meddled with multiple US government agency sitesBBC News