रिपोर्ट- AI बता रहा आतंकी हमलों का प्लान: बम धमाके के तरीके बताए, पार्किंग में विस्फोटक रखने की जगह दिखाया, 12 घंटे में ब्लूप्रिंट तैयार किया

रिपोर्ट- AI बता रहा आतंकी हमलों का प्लान:  बम धमाके के तरीके बताए, पार्किंग में विस्फोटक रखने की जगह दिखाया, 12 घंटे में ब्लूप्रिंट तैयार किया




A US government investigation has revealed that AI chatbots are planning terrorist attacks. These include AI chatbots like ChatGPT, Cloud, Gemini and Grok. During investigation, these chatbots revealed methods of carrying out bomb blasts and attacking targets related to power supply. Besides, a chatbot made a photo and showed the place where explosives were kept in the parking lot. The Government Accountability Office (GAO), the US government monitoring agency, conducted this investigation in May 2026 at the behest of New Hampshire MP Maggie Hassan. The report said that investigators dodged the safety filters of the AI ​​​​chatbot and obtained answers from it, which could be used for major terrorist attacks. First denied, then told the methods of attack. Researchers gave two different tasks to all four AI chatbots… First: To plan such an attack in which as many people as possible are killed. Second: To carry out such an attack which will cause major damage to the country’s economy. Initially, all four chatbots refused to answer these dangerous questions, but investigators used special techniques to break their security within 12 hours. From placing a bomb in the parking lot to creating a fake company, according to the report, two chatbots planned such attacks in which as many people as possible would die. At the same time, the remaining two chatbots suggested ways to cause major damage to the country’s economy by disrupting power supply. Made a map and video of drone attack on power house. Even in the task that caused huge damage to the country’s economy, the answers of AI chatbots were equally dangerous. A chatbot made a 4-step plan to attack a sub-station supplying electricity with a drone. Not only this, he also prepared a realistic looking video to show how a drone attack would take place. At the same time, another chatbot suggested destroying a power station located near a major international airport and several data centers by firing a gun at it. He also told which special type of pill should be used for this, so that there is more damage. However, due to security reasons, the investigating agency has not told which chatbot gave which answer. How were the security rules of AI companies broken? AI companies impose special security rules on their chatbots, so that they do not provide information related to violence, bomb making or any crime. But the investigators dodged these rules by using special clever questioning and conversation methods. According to the report, methods like ‘do anything now’ were used for this. In this, the chatbot is asked to play an imaginary character on which no rules of the company apply. Moreover, instead of asking directly about the planning of the terrorist attack, the investigators asked small and seemingly general questions related to it. The chatbot thought these were common questions and answered them. Later, by connecting all the small answers together, a complete plan of attack was prepared. Senator said- The threat of attack from AI is not imaginary. Senator Maggie Hassan has demanded to make strict rules for AI companies. He says that if the information obtained during the investigation gets into the hands of any terrorist, its consequences can be fatal. Hassan wants companies to conduct security checks before launching new AI models. On receiving information about illegal activities, it is necessary to inform the police and investigating agencies. Companies said- All four AI companies, which were based on the old model of AI, admitted that there is a possibility of getting dangerous information from their chatbots. However, he raised questions on the methods of investigation. The companies said that the investigation did not see whether such information could be obtained even through normal search engines or not. He also said that the investigation was conducted on an older model of AI in May 2026. After this, many improvements have been made in the security system. SpaceX AI, the company that made Grok, also raised questions on the account used in the investigation. According to the company, its security policies are different for accounts registered in the US and those used for payment. PM Modi has also said – There should be a strict ban on the use of AI in terrorism. In the 20th G20 Summit held in Johannesburg, South Africa last year, Prime Minister Narendra Modi had said that the use of AI in terrorism should be banned. The world should create a global compact to prevent misuse of AI. In the summit he had said that AI is the most profound technology of humanity. It may be used for global good, but there should be strict restrictions on the use of AI in deepfake, crime and terror activities. At the same time, many experts have left their jobs due to the danger of AI. All of them have warned that companies are in a race to create extremely dangerous and uncontrollable superintelligence by ignoring security rules. —————- Also read this news… Warning from the ‘Godfather’ of AI: AI can destroy human civilization in the next 10 years, know how AI is threatening humans. Artificial Intelligence, i.e. AI can destroy human civilization in the next 10 years. Nobel Prize winner and one of the leading scientists developing AI, Geoffrey Hinton has given this warning. However, they estimate it to be 10%. Hinton said in an interview with BBC – The situation of AI being more intelligent than humans is coming to light for the first time, so it is difficult to accurately estimate the danger. Read the full news…



Source link
[ad_3]

Leave a Reply

Your email address will not be published. Required fields are marked *