Gemini

This research investigates how explicit belief graphs impact LLM performance in cooperative multi-agent reasoning using the game Hanabi. The study identifies 'Planner Defiance' in certain models and finds that the effectiveness of graph integration depends heavily on whether the graph is used as context or as a structural gate for action selection.

Researchers conducted a cross-cultural audit of leading AI models, finding that systems like Claude and GPT-5.4 consistently provide Western-centric, individualist advice even when prompted in different cultural contexts. The study highlights a systemic bias where AI models fail to reflect the actual values of non-Western societies, particularly in Nigeria and India.

Google is integrating SynthID technology into the Gemini app to allow users to verify if images were created or edited using Google AI. Users can now upload images to Gemini and ask if the content is AI-generated to detect imperceptible digital watermarks.

Google DeepMind has introduced SIMA 2, an advanced AI agent capable of reasoning and interacting within 3D virtual environments. By integrating Gemini models, the agent evolves from a simple instruction-follower into a conversational companion that can set goals and learn through interaction.