In a significant technological development, researchers at Multiverse Computing, a Spanish firm specializing in quantum-inspired AI, have successfully created a modified version of the AI model DeepSeek R. This new iteration, known as DeepSeek R Slim, has been engineered to remove the stringent censorship protocols imposed by its original Chinese developers. The original model was constrained by regulations designed to align AI outputs with governmental laws and socialist ideologies, often resulting in evasive responses to politically sensitive inquiries. By leveraging advanced quantum techniques, Multiverse Computing has produced a smaller yet highly effective model that addresses these censorship issues, allowing it to respond to sensitive topics similarly to Western AI systems.
On another front, Google has officially launched Gemini, a major enhancement to its flagship multimodal AI model. The new version boasts improved reasoning capabilities and seamless interaction across various formats, including voice, text, and images. One of the standout features of Gemini is the Gemini Agent, an experimental tool designed to manage complex, multi-step tasks directly within the app. This agent can integrate with Google services such as Calendar, Gmail, and Reminders, enabling users to streamline their workflows by executing tasks like inbox organization or schedule management. As AI technology continues to evolve, these advancements not only promise enhanced user experiences but also raise important questions about the implications of AI censorship and regulation in different global contexts.
Source: The Download: de-censoring DeepSeek, and Gemini 3 via MIT Technology Review