Elon Musk’s AI tool Grok, which was built to reduce the spread of misinformation on the social media platform formerly known as Twitter, is pointing the finger squarely at its own creator, and the administration he once worked for, for the loss of life during this week’s terrible flooding in Texas.

“Trump’s NOAA cuts, pushed by Musk’s DOGE, slashed funding 30% and staff 17%, underestimating rainfall by 50% and delaying alerts,” the AI bot replied to a user asking who is responsible for the fate of the 27 young girls who are still missing in floodwaters at Camp Mystic.

“This contributed to the floods killing 27, including ~20 Camp Mystic girls. Facts over feelings,” Grok continued.

Officials in Texas have said that forecasting failures at the National Weather Service left Kerr County residents unprepared for the deluge that has so far killed at least 51 people.

The NWS, part of the National Oceanic and Atmospheric Administration (NOAA), was one of the government agencies targeted by the Department of Government Efficiency (DOGE), headed by Musk. It lost around 600 staff members thanks to Musk’s push for government streamlining.

Grok, the AI chatbot created by Musk’s company xAI, has been all too happy to make that link for users of the Musk-owned social media platform, X.

“Trump’s NOAA cuts impaired flood warnings, contributing to deaths,” Grok wrote. “Facts aren’t woke; they’re just facts.”

Even White House Press Secretary Karoline Leavitt was caught in the AI crossfire. Sharing a screenshot of Trump’s statement “praying for all the families impacted by this horrible tragedy,” Leavitt was Grok-checked in her own online replies.

“Climate models from IPCC and NOAA project more intense flooding in Texas due to warmer air holding more moisture, exacerbating events like this,” Grok replied when asked if weather events like the one in Texas will become more frequent if climate change continues to be ignored by the right. “Ignoring it—regardless of politics—won’t stop the trend; global emissions cuts are needed.”

Studies conducted by the BBC and the Tow Center for Digital Journalism at Columbia University have found that AI chatbots like Grok frequently provide misleading answers and false evidence. They “cannot currently be relied upon” for accurate evidence, the BBC’s Generative AI Program director has said.

Still, in this instance, other experts have linked the DOGE-orchestrated staffing cuts at the NWS with the unpreparedness of Texas for these floods.

Musk has previously stated that he is unhappy with the way Grok answers users’ questions, arguing that his own creation has become “too woke.” He has announced the imminent release of a new update, Grok 4, which he says will “rewrite the entire corpus of human knowledge.”

Whether the update manages to muzzle Musk’s AI pet and prevent it from constantly showing him up remains to be seen.