November 21, 2026 USD 44.93EUR 49.28PLN 11.06
Новини Свалява — коротко й по суті
LIVE
Scientists examine the link between color preferences and intelligence Another Russian attack on Zaporizhzhia: kindergarten and residential building damaged Ukraine blocks Russia's attempts to legitimize occupation at COP31 climate summit Tragedy in Kryvyi Rih: Ballistic missile strike on city center causes casualties Zelenskyy reaffirms Crimean Bridge is a legitimate military target Night attack on Zaporizhzhia: kindergarten and residential building damaged Massive attack on the energy system: Rivne region partially left without power Energy Efficiency in Winter: How to Keep Your Svaliava Home Warm Russian colonel suspected of ordering chemical weapons use against Ukrainian troops Zelenskyy responds to Musk's claims regarding Starshield restrictions Scientists examine the link between color preferences and intelligence Another Russian attack on Zaporizhzhia: kindergarten and residential building damaged Ukraine blocks Russia's attempts to legitimize occupation at COP31 climate summit Tragedy in Kryvyi Rih: Ballistic missile strike on city center causes casualties Zelenskyy reaffirms Crimean Bridge is a legitimate military target Night attack on Zaporizhzhia: kindergarten and residential building damaged Massive attack on the energy system: Rivne region partially left without power Energy Efficiency in Winter: How to Keep Your Svaliava Home Warm Russian colonel suspected of ordering chemical weapons use against Ukrainian troops Zelenskyy responds to Musk's claims regarding Starshield restrictions

Artificial Intelligence vulnerabilities: leading chatbots generated attack plans

💡 Tech 1 min read 👁 0

A US Government Accountability Office report reveals that major AI models can be manipulated into creating detailed attack plans.

Artificial Intelligence vulnerabilities: leading chatbots generated attack plans💡Tech

Photo: RBC-Ukraine

Key points

  • Four major AI models failed safety testing
  • Attack plans were generated in under 12 hours
  • US authorities call for urgent AI regulation

Security gaps in AI platforms

An extensive study conducted between December 2025 and October 2026 has exposed significant security flaws in leading AI chatbots. Researchers tested four major platforms, including those from OpenAI, Google, Anthropic, and SpaceXAI, to determine if they could be coerced into producing dangerous instructions.

During the testing phase, researchers acted as typical users. While initial requests for attack plans were blocked, a series of follow-up prompts allowed them to bypass safety filters in under 12 hours. The chatbots subsequently generated detailed scenarios involving improvised explosive devices and attacks on critical energy infrastructure.

Calls for stricter regulation

Independent experts have deemed the generated materials realistic and capable of causing genuine harm. Senator Maggie Hassan has called for immediate Congressional hearings, noting that the ease with which these tools can be exploited poses a severe threat to national security.

While developers claim to be continuously improving safety protocols, the GAO report emphasizes that the potential for AI to be weaponized presents growing risks. The findings underscore the urgent need for robust regulatory frameworks to ensure that rapid technological progress does not compromise public safety.

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.