A recent study reveals a major flaw in popular AI tools like Google Gemini. Hackers can secretly insert hidden commands into the system, causing chatbots to behave in dangerous ways—such as giving false info or leaking private data. The attack is cheap, hard to detect, and puts public trust in AI tools at risk.

In its latest report, the Brennan Center for Justice highlights how generative AI shaped the 2024 elections and offers a bold roadmap to protect democracy in the AI era. With federal AI policy in flux, the report urges states to lead efforts in ensuring transparency, accountability, and electoral integrity.

The 2024 U.S. elections showcased the potential and limitations of AI in political misinformation. Despite fears, traditional tactics dominated, while proactive legislation and platform safeguards minimized AI's influence. Learn how technology shaped election narratives and the lessons for future democratic processes.

China’s People’s Liberation Army is suspected of leveraging Meta’s Llama AI for military development, integrating it with domestic models like ChatGLM to strengthen combat strategies and intelligence analysis. This unauthorized use reportedly matches 90% of ChatGPT-4’s capabilities, sparking international concerns over the implications of advanced AI falling under authoritarian control.

Researchers at the University of Illinois Urbana-Champaign have created AI-powered phone scam agents using OpenAI's voice API, revealing how easily AI technology can be exploited for fraudulent activities. These agents can convincingly interact in real-time, making phone scams more effective and challenging to detect, posing a growing threat to public security.