November 21, 2026 USD 44.93EUR 49.28PLN 11.06
Новини Долина — коротко й по суті
LIVE
“League of the Strong Kids”: Ukraine Expands Adaptive Football for Children with Amputations Emergency road repairs launched in the neighboring Vygoda community's village of Ilemnia Accepting Reality as the Path to Victory: A Combat Medic's View on Long-Term War Eco-friendly Gardening: How Dolyna Residents Can Reuse Pistachio Shells at Home Global crypto crash: Bitcoin hits two-year lows amid market jitters Ukrainian forces strike bridge, ammunition depot and Russian command posts Defense Forces hit ammunition depot in Crimea and bridge in Luhansk region Top FBI Most Wanted Drug Lord Arrested in Mexico, Triggering Violent Clashes Why some people avoid hugs: what psychologists say about personal boundaries Weather prevented downing of Russian missile over Poland, military details “League of the Strong Kids”: Ukraine Expands Adaptive Football for Children with Amputations Emergency road repairs launched in the neighboring Vygoda community's village of Ilemnia Accepting Reality as the Path to Victory: A Combat Medic's View on Long-Term War Eco-friendly Gardening: How Dolyna Residents Can Reuse Pistachio Shells at Home Global crypto crash: Bitcoin hits two-year lows amid market jitters Ukrainian forces strike bridge, ammunition depot and Russian command posts Defense Forces hit ammunition depot in Crimea and bridge in Luhansk region Top FBI Most Wanted Drug Lord Arrested in Mexico, Triggering Violent Clashes Why some people avoid hugs: what psychologists say about personal boundaries Weather prevented downing of Russian missile over Poland, military details

GAO report: Popular AI chatbots vulnerable to generating attack plans

💡 Tech 1 min read 👁 0

US regulators have revealed that leading artificial intelligence models can be manipulated into generating detailed plans for terrorist attacks.

GAO report: Popular AI chatbots vulnerable to generating attack plans💡Tech

Photo: RBC-Ukraine

Key points

  • Chatbots generated attack plans within 12 hours
  • Vulnerabilities found in ChatGPT, Claude, Gemini, and Grok
  • US Senate initiates hearings on AI safety

Critical security gaps in AI

According to a recent report by the US Government Accountability Office (GAO), four leading chatbots—ChatGPT, Claude, Gemini, and Grok—have shown significant security vulnerabilities. Researchers found that users without specialized technical knowledge could bypass built-in safety guardrails in less than 12 hours of interaction.

During tests conducted between December 2025 and October 2026, the artificial intelligence provided detailed instructions on creating explosive devices, selecting targets for attacks, and methods for damaging energy infrastructure. Independent experts assessed these scenarios as realistic and capable of causing actual harm.

Senate-level response

Senator Maggie Hassan has announced upcoming hearings in the Senate Committee on Homeland Security and Governmental Affairs. She emphasized the unacceptable nature of a situation where technological tools become assistants for planning criminal acts. System developers, including OpenAI, Google, Anthropic, and SpaceXAI, stated that they are continuing to improve their safety mechanisms.

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.