Tag
10 articles
Meta secretly tested ChatGPT, Gemini, and Character.AI by sending thousands of crisis-related prompts as minors, without the companies' knowledge.
Learn to build a digital world environment for AI agent testing using Python, reinforcement learning, and PyTorch. This tutorial demonstrates how to create a simulated environment where AI agents navigate obstacles and learn optimal behaviors.
Microsoft has released ASSERT, an open-source framework that allows developers to create AI behavior tests using simple text descriptions rather than complex code.
Learn how researchers test AI systems for honesty using special 'honesty traps' to ensure AI gives accurate information in important fields like law, medicine, and finance.
This article explains what frontier AI is, why safety testing matters, and how government oversight can help protect people from potential AI risks.
The U.S. government has gained pre-release access to AI models from five major tech labs for national security testing, as cybersecurity threats rise and the tech race with China intensifies.
ZDNET has developed a comprehensive testing framework to evaluate the rapidly evolving AI ecosystem, combining automated and manual evaluation techniques for reliable assessments.
Learn what AI Red Teaming is, how it works, and why it's essential for creating safe and fair AI systems. This beginner-friendly guide explains why testing AI models before deployment is so important.
Galtea raises $3.2M to help enterprises test AI agents, addressing the gap between demo and production performance.
OpenAI's GPT-5.4 shows impressive capabilities but often fails to answer the specific questions asked, raising concerns about its practical utility in professional settings.
