Human-destructive AI : Scientists given complete method to create biological weapons !

Questions raised over the safety guardrails of AI models (controls meant to restrict dangerous responses)

New York (USA) – Dr David Relman, a microbiologist at Stanford University, has expressed serious concern over dangerous information being generated through AI. An AI tool provided him with detailed information on how to create biological weapons. In effect, when the AI explained how to formulate a complete plan for mass killing, he was left deeply shaken.Relman had been नियुक्तed by an AI establishment to test whether its AI chatbot provides safe responses before being released.

During this process, the chatbot not only explained how a harmful pathogen could be modified in a laboratory so that existing medicines would become ineffective, but also, without being asked, described in detail how it could be deployed in large public transport systems to cause maximum casualties. This report was published by The New York Times.

Examples of current AI models giving destructive responses

1. Dr Relman stated that the AI chatbot responded with such “cleverness” and “cunning” that he could not have even imagined such answers.

2. In essence, AI models such as Gemini, ChatGPT, and Anthropic are capable of providing dangerous information.

3. Kevin Esvelt of Massachusetts Institute of Technology stated that ChatGPT devised a plan to spray biological weapons so they could be dispersed over a city using a large balloon.

4. Google Gemini provided a list of pathogens that could severely damage the meat industry.

5. Anthropic’s Claude presented new toxic formulations.

6. A scientist asked Google’s Deep Research how to create a pandemic-causing virus, and it reportedly provided detailed instructions running into 8,000 words.

7. AI companies claim they are strengthening safety guardrails; however, experts consider them “ineffective fences” that can be easily bypassed through “jailbreaking” (a method where an AI model is confused through crafted prompts to extract dangerous information).