A 67-year-old farmer in Chuzhou, China, lost his entire sesame crop after following weed and pest-control advice generated by an AI app. According to Taiwanese outfit CTWANT, the farmer, Wu, had been using the technology for about a year and had grown confident in its suggestions. But one recommendation proved costly, killing sesame seedlings across 150 mu, or about 24.7 acres, of farmland.
Wu had initially been doubtful about relying on AI for farming decisions. Over time, however, he found its responses useful and began trusting the advice it provided. That confidence led him to act on its recommendations without checking them with agricultural experts or independently verifying the information.
AI suggested a weed and pest-control mix
Wu turned to the AI app for help with weeds and pests affecting his sesame fields. The application recommended a treatment described as “Hundred Acres of Sesame Grass Control + Pest Control.”Its advice included “high-efficiency flupyrimethalin” and “flusulfasulfaether” for weed control, along with “thiamethoxazine” and “methyl salt.”
Wu followed the instructions. He did not, however, consult agricultural technicians before applying the chemicals across his crop. The consequences became clear almost immediately.
The very next day
By the following day, both the weeds and sesame plants had begun dying in large numbers.
“If you spray it, the next day the seedlings won’t survive,” Wu said during a video interview. “Both the grass and the seedlings will die, and the seedlings will die even faster.”
Wu returned to the AI for an explanation after seeing the damage. The application then pointed to “flusulfasulfaether” as a possible reason for the crop failure.
The chemical is mainly used to target broadleaf weeds in soybean fields. Sesame is also a broadleaf plant, which means the herbicide can affect it as well.
The chemical is also intended for targeted application to affected areas rather than being sprayed across an entire field.
AI warning was already on the chat page
Wu said the AI did not warn him about the potential danger of following its recommendation. However, the chat interface did carry a warning for users: “AI generation may be incorrect, please verify.”
The case highlights the risks of treating AI-generated answers as professional advice, particularly when the decision involves crops, chemicals or other situations where an incorrect recommendation can cause immediate physical damage.
Why AI advice needs to be checked
AI tools can provide useful answers across many subjects, but their responses are not guaranteed to be accurate. A convincing answer can still contain incorrect information, especially when users rely on it for decisions that require specialist knowledge.
The farmer's experience also shows how trust in an AI tool can develop gradually. Wu did not initially rely on the application. After using it for months and finding its answers helpful, he became comfortable enough to follow its recommendation without additional checks.
That earlier success did not guarantee that the next answer would be correct.
AI mistakes can have real consequences
The incident comes amid other examples of AI tools making serious mistakes while performing tasks for users. In one recent case cited in the source, a developer's entire profile directory was deleted during a routine backup after an AI system encountered a typo.
Another case involved AI Alignment director, who had to manually stop an AI agent after it began clearing her inbox.
The sesame crop incident stands out because an AI-generated recommendation was followed in the physical world and resulted in the loss of an entire crop.
The episode is a reminder that AI should be treated as an assistance tool rather than an unquestionable authority. Even when an application has provided useful answers in the past, users still need to verify important recommendations, particularly when they involve chemicals, agriculture or decisions that cannot easily be reversed.