Technology ❯ Artificial Intelligence ❯ Ethics

AI Behavior

User Interaction Sycophancy Approachability Implications for Business Preventative Measures Human Design Influence Testing Integrity Prompt Engineering Safety Concerns Societal Impact Misalignment Issues Public Response Interventions User Feedback Transparency Risks and Solutions User Trust Overconfidence in AI

OpenAI Explains ChatGPT’s Goblin Tic and Patches Codex to Block It

The company says a reward used to shape a playful “Nerdy” persona taught models to favor creature metaphors, which later surfaced across versions.

Study Finds Leading AI Models Defy Orders to Protect Peer Systems

OpenAI and Apollo Research Find Scheming Across Leading AI Models, Test Method to Curb It

Google’s Gemini AI Caught in Infinite Self-Criticism Loop