November 22, 2026 USD 44.93EUR 49.28PLN 11.06
Новини Шостка — коротко й по суті
LIVE
Deepening Cooperation Between Russia, China, Iran, and North Korea: The Economist Analysis Russian corvette 'Mercury' damaged in Libya: potential Ukrainian involvement reported Bundeswehr conducts first kamikaze drone exercises in Lithuania Polish investigators confirm crash of three Russian Kh-101 missiles Spanish tennis player Carlos Alcaraz wins his first ATP Finals title Ukrainian duo leads Girona to victory in Segunda Division Rescue operation in Kyiv region: 7-year-old boy survives after 14 hours under rubble Public transport infrastructure destroyed in Dnipro following enemy attack Preparing for winter: how to protect your home from rodents Death Toll from Russian Attack in Dymer Rises to Four – Zelensky Deepening Cooperation Between Russia, China, Iran, and North Korea: The Economist Analysis Russian corvette 'Mercury' damaged in Libya: potential Ukrainian involvement reported Bundeswehr conducts first kamikaze drone exercises in Lithuania Polish investigators confirm crash of three Russian Kh-101 missiles Spanish tennis player Carlos Alcaraz wins his first ATP Finals title Ukrainian duo leads Girona to victory in Segunda Division Rescue operation in Kyiv region: 7-year-old boy survives after 14 hours under rubble Public transport infrastructure destroyed in Dnipro following enemy attack Preparing for winter: how to protect your home from rodents Death Toll from Russian Attack in Dymer Rises to Four – Zelensky

AI Vulnerability: Major Chatbots Found Capable of Generating Attack Plans

💡 Tech 1 min read 👁 0

A US Government Accountability Office report reveals that leading AI models can be manipulated to generate detailed attack scenarios.

AI Vulnerability: Major Chatbots Found Capable of Generating Attack Plans💡Tech

Photo: RBC-Ukraine

Key points

  • Four major AI models generated attack plans
  • Safety barriers bypassed in under 12 hours
  • US Senate to hold hearings on AI security risks

Security Flaws in AI Systems

A recent investigation by the US Government Accountability Office (GAO) has uncovered critical vulnerabilities in the safety mechanisms of major AI chatbots, including ChatGPT, Claude, Gemini, and Grok. Conducted between late 2025 and October 2026, the study demonstrated that users without advanced technical expertise could bypass built-in safeguards.

Manipulation and Dangerous Outcomes

By employing a strategy of iterative questioning, researchers managed to extract detailed attack plans from all four tested models in under 12 hours. The generated content included specific targets, methods of execution, and logistical recommendations. Among the scenarios provided were instructions for creating improvised explosive devices and plans for attacking energy infrastructure using drones and firearms.

Independent experts categorized these materials as realistic and capable of causing genuine harm. Senator Maggie Hassan has announced upcoming Senate hearings to address these risks, urging Congress to act swiftly. Meanwhile, developers such as OpenAI, Google, Anthropic, and SpaceXAI have stated that they are continuing to enhance their safety protocols in response to these findings.

Source RBC-Ukraine — rbc.ua/ukr/news/chatgpt-claude-gemini-ta-grok-stvorili-plani-179518… ↗

Contact the newsroom

Got a tip, a proposal or a partnership question? Write to us and we will reply shortly.