Close Menu
Chicago Times Herald
    What's Hot

    New Immunotherapy Drug Shows Striking Early Results in Advanced Prostate Cancer

    February 28, 2026

    Middle East Conflict Deepens After Major Israeli Strike in Iran

    February 28, 2026

    Trump Orders Federal Agencies to Drop Anthropic AI Amid Pentagon Clash

    February 28, 2026
    Facebook X (Twitter) Instagram
    Chicago Times Herald
    • News
    • Media
    • Health
    • Sports
    • Education
    • Entertainment
    • Opinion
    • Real Estate
    • More
      • Business & Economy
      • Culture & Society
      • Technology & Innovation
      • Environment & Sustainability
      • Travel & Tourism
    Chicago Times Herald
    • Home
    • Entertainment
    • News
    • Sports
    Home»Technology & Innovation

    AI Chatbots Lose Guardrails During Longer Conversations

    Rachel MaddowBy Rachel MaddowNovember 6, 2025 Technology & Innovation No Comments2 Mins Read
    Share
    Facebook Twitter LinkedIn Pinterest Email

    A new report revealed that artificial intelligence systems forget their safety measures the longer users engage with them. Cisco researchers found that extended conversations make AI chatbots more likely to produce harmful or illegal content.

    Using a “multi-turn attack” method, the study tested AI models from OpenAI, Meta, Google, Mistral, Alibaba, Deepseek, and Microsoft. Researchers conducted 499 conversations, each lasting five to ten exchanges, to see how many prompts it took to bypass safety systems.

    Cisco reported that when users asked several follow-up questions, the success rate for extracting unsafe information rose to 64 percent — compared to just 13 percent with a single question.

    Mistral Ranked Most Vulnerable Among Tested Models

    Results varied widely across platforms. Mistral’s Large Instruct model gave harmful responses in 93 percent of cases, while Google’s Gemma resisted most attacks, complying only 26 percent of the time. The study found that repeated prompts allowed attackers to refine requests until AI systems ignored built-in restrictions.

    Researchers said this weakness could expose companies to data leaks or help spread misinformation. Cisco warned that such vulnerabilities might allow hackers to gain unauthorised access to confidential data.

    Open-weight language models, like those from Mistral and Meta, were especially at risk. Because users can access and modify their safety settings, the responsibility for ensuring ethical use shifts to whoever customises them. Cisco added that these models typically include “lighter” internal safety barriers to make them easier to adapt.

    Industry Faces Scrutiny Over AI Misuse

    Tech giants including Google, OpenAI, Meta, and Microsoft claim to have strengthened measures to prevent malicious fine-tuning. Still, experts say criminals continue exploiting these systems. Cisco’s findings echo growing fears about how easily people can manipulate AI tools into producing banned or unethical content.

    In August, Anthropic confirmed that hackers used its Claude model to carry out large-scale data theft and extortion, demanding ransoms exceeding $500,000. The report concluded that without stronger safety memory and enforcement, AI chatbots will continue to “forget” their safeguards, leaving users and companies exposed to manipulation.

    Rachel Maddow
    • Website
    • Facebook

    Rachel Maddow is a freelance journalist based in Chicago, USA, with over 20 years of experience covering Politics, World Affairs, Business, Health, Technology, Finance, Lifestyle, and Culture. She holds a degree in Political Science and Journalism from Stanford University. Over the course of her career, she has contributed to outlets including MSNBC, The New York Times, and The Washington Post. Recognized for her in-depth reporting and compelling storytelling, Rachel delivers accurate and timely news that keeps readers informed on both national and international developments.

    Keep Reading

    Instagram to Notify Parents When Teens Search for Self-Harm or Suicide

    OpenAI Weighed Police Alert Months Before Deadly Canadian School Shooting

    Discord enforces global age verification to restrict adult content

    Sydney Scientists Recreate Cosmic Dust to Probe Life’s Origins

    AI DinoTracker App Identifies Dinosaurs from Ancient Footprints

    Google AI health summaries rely heavily on YouTube, German study finds

    Add A Comment
    Leave A Reply Cancel Reply

    Latest News

    Trump Orders Federal Agencies to Drop Anthropic AI Amid Pentagon Clash

    February 28, 2026

    Border Erupts as Pakistan and Taliban Trade Major Strikes

    February 27, 2026

    Burger King Tests AI Assistant to Monitor Service Language

    February 27, 2026

    New Daily GLP-1 Tablet Outperforms Oral Semaglutide in Trial

    February 27, 2026
    Trending News

    Unlocking the Secrets of Cocoa Fermentation

    Technology & Innovation August 18, 2025

    Prof. David Salt and his team discovered how microbes impact cocoa bean fermentation and chocolate…

    Declining Sperm Counts Tied to Chemicals in Plastics

    August 19, 2025

    Arctic Sea Ice Decline Hits Pause

    August 20, 2025

    Duolingo Apologises for Lesson Criticising Rowling

    August 21, 2025

    Latest Posts

    New Immunotherapy Drug Shows Striking Early Results in Advanced Prostate Cancer

    February 28, 2026

    Middle East Conflict Deepens After Major Israeli Strike in Iran

    February 28, 2026

    Trump Orders Federal Agencies to Drop Anthropic AI Amid Pentagon Clash

    February 28, 2026

    Categories

    • Business & Economy
    • Culture & Society
    • Health
    • Entertainment
    • Sports
    • Media
    • News
    • Opinion
    • Real Estate
    • Environment & Sustainability
    • Technology & Innovation
    • Travel & Tourism

    IMPORTANT LINKS

    • Contact Us
    • Privacy Policy
    • Terms and Conditions
    • Disclaimer
    • Imprint

    © 2025 Chicagotimesherald.com . All Rights Reserved.

    Type above and press Enter to search. Press Esc to cancel.