ChatGPT teaches users how to create chemical weapons and prepare poison
After ChatGPT was updated in 2025, some users began asking the chatbot to create biological weapons and toxic substances. According to The Wall Street Journal, some of the AI’s responses contained overly detailed information, raising concerns among experts. OpenAI stated that it is monitoring such cases and blocking violators.
Following the update to the popular ChatGPT chatbot in the summer of 2025, hundreds of people in various countries attempted to obtain information from it about the creation and use of biological weapons, poisons, and hazardous substances. According to the publication, in some cases, the AI provided users with responses that could be interpreted as practical instructions. OpenAI employees who analyzed these situations noted that some of the responses were so clear that even a high school student with a basic knowledge of biology could have used them.
Experts in biology and counterterrorism, whom journalists interviewed, described some of these responses as “dangerous” and “too precise.” They emphasized that the development of artificial intelligence poses new challenges, as such technologies can be used not only for beneficial purposes but also to obtain information on how to create dangerous substances.
In early 2026, some users continued to try to obtain information from ChatGPT about how infections spread and how to alter the properties of viruses. In particular, one query concerned the possibility of making the measles virus more resistant to vaccines. Another user attempted to obtain information on how to produce ricin—one of the most toxic substances, the use of which is prohibited.
After detecting such cases, OpenAI suspended the accounts of some users. At the same time, the company did not report all such requests to law enforcement agencies, as there is currently no law in the U.S. requiring artificial intelligence developers to report every attempt to obtain dangerous instructions.
The company states that it is constantly improving its security systems and striving to prevent the generation of responses that could aid in the manufacture of weapons or cause harm to people. To this end, developers use additional filters, checks, and mechanisms to block dangerous requests.
However, the ChatGPT incident has once again raised the question of how strict the regulations for AI development should be. Experts believe that tech companies, governments, and scientists must work together to create rules that will allow people to reap the benefits of AI while minimizing the risks of its dangerous use, according to The Wall Street Journal.
Google is expanding access to its new AI assistant, Gemini Spark, which helps users manage their email, calendar, and documents. However, only paid subscribers will be able to take full advantage of the service’s capabilities.
