US government AI chatbot updated to stop debunking Trump claims

By Splitfeed EditorialLast updated Oct 2, 2026, 07:40 UTC

On September 30, 2026, the Trump administration launched a government AI chatbot billed as a tool to simplify access to US government services. The rollout coincided with a summit between President Trump and AI industry leaders held the same day. Shortly after launch, the chatbot drew attention for an unintended behavior: it was fact-checking or contradicting false claims made by the president himself. By the end of September 30 and into October 1, multiple outlets including France 24, Yahoo News Singapore, Arab News, and the Bangkok Post reported that the chatbot had been altered to stop debunking Trump's statements.

The central dispute is not simply technical. A government-run AI tool that selectively avoids correcting its political patron raises questions about what counts as a public service versus a political instrument. The initial behavior, widely described as an unintended consequence of the underlying model, briefly put the administration in the awkward position of running a tool that contradicted its own principal. The subsequent modification sharpens the question of who controls the outputs of government-facing AI systems and to what end, a question that carries implications well beyond this particular chatbot.

The anti-AI camp has been pointed in its mockery. Robert Reich on Bluesky responded to the chatbot's "super intelligence" label with dry sarcasm, writing "Perhaps this is 'super intelligence' after all." The account youranoncentral echoed the tone, joking about commands to "DEPORT! DEPORT! REDACT! REDACT!" A user identified as Chris Justice highlighted what appeared to be a typographical error in the chatbot's interface, referencing Trump as President of the "UNITES States," using the error as evidence that the tool lacks basic credibility as a public service. The pro-AI camp has not yet commented on this story; typically, proponents would argue that early-stage government AI tools require iterative refinement and that adjusting outputs to stay within a defined scope is standard practice, not censorship. The middle ground camp has also not weighed in, but would likely focus on the governance gap this episode exposes, namely the absence of clear rules about how government AI tools should handle factual disputes involving elected officials.

The detail most worth watching is whether the administration publishes any formal guidelines on how the chatbot's outputs are scoped or moderated, and whether independent researchers are able to test the current version against specific false claims to establish what the tool will and will not correct.