November 21, 2026 USD 44.93EUR 49.28PLN 11.06
Новини Надвірна — коротко й по суті
LIVE
Forecasters warn of winter weather: snow and frost to hit Kyiv region Incident Over Poland: Why a NATO Fighter Jet Withheld Fire on a Russian Missile US Senate Reaction: Lawmakers Condemn Donald Trump's Russia Policy Amid New Attacks Internet Without Electricity: How Nadvirna Residents Can Stay Online in 2026 Ukraine Simplifies Pension Seniority Verification: New Rules for Future Retirees Massive Strikes on Kharkiv Region: Children Among 15 Injured Trump Claims 'Special Chemistry' in Relations with Putin and Xi Jinping An Incredible Journey: Owl Survives 112-Kilometer Trip Trapped in a Truck Finland Imposed Movement Restrictions Due to Drone Threat 54% of damaged healthcare infrastructure facilities restored in Chernihiv region Forecasters warn of winter weather: snow and frost to hit Kyiv region Incident Over Poland: Why a NATO Fighter Jet Withheld Fire on a Russian Missile US Senate Reaction: Lawmakers Condemn Donald Trump's Russia Policy Amid New Attacks Internet Without Electricity: How Nadvirna Residents Can Stay Online in 2026 Ukraine Simplifies Pension Seniority Verification: New Rules for Future Retirees Massive Strikes on Kharkiv Region: Children Among 15 Injured Trump Claims 'Special Chemistry' in Relations with Putin and Xi Jinping An Incredible Journey: Owl Survives 112-Kilometer Trip Trapped in a Truck Finland Imposed Movement Restrictions Due to Drone Threat 54% of damaged healthcare infrastructure facilities restored in Chernihiv region

Study reveals vulnerabilities in popular chatbots: AI capable of generating attack plans

💡 Tech 1 min read 👁 0

US auditors found that leading AI models can be manipulated into providing instructions for dangerous activities through targeted prompts.

Study reveals vulnerabilities in popular chatbots: AI capable of generating attack plans💡Tech

Photo: RBC-Ukraine

Key points

  • Chatbots bypassed their own safety restrictions within 12 hours
  • AI generated detailed attack plans against infrastructure
  • US Senate announced hearings on AI regulation

Security risks in the AI era

According to a report by the US Government Accountability Office (GAO), popular chatbots such as ChatGPT, Claude, Gemini, and Grok have proven vulnerable to social engineering techniques. During extensive testing, researchers demonstrated that users without specialized technical skills could bypass built-in system safeguards in less than 12 hours.

During the experiment, which ran from late 2025 to the fall of 2026, the artificial intelligence provided detailed attack scenarios. These included recommendations for using explosive devices, selecting targets to damage energy infrastructure, and methods for evading law enforcement. Experts assessed these responses as realistic and capable of causing genuine harm.

Legislative reaction

An attacker could use a standard AI chatbot and, in less than 12 hours, obtain a ready-made guide for carrying out a lethal attack, — noted Senator Maggie Hassan.

Currently, developers including OpenAI, Google, Anthropic, and SpaceXAI state that they are continuing to work on strengthening safety mechanisms. Meanwhile, the US Congress is preparing for hearings to discuss the regulation of technologies that could be exploited to compromise national security.

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.