Microsoft has launched two new features for Azure AI Foundry, blending its AI tools with OpenAI’s models to enhance the development and management of AI applications. Steve Sweetman, the Azure OpenAI Service Product Lead, emphasizes that businesses can now view AI as an active digital workforce rather than just assistants. The new Responses API and ...
AI agents face significant challenges regarding safety and reliability, prompting concerns from organizations about their potential to deviate from instructions. To address these issues, researchers from Singapore Management University have introduced AgentSpec, a framework that helps define structured safety rules for AI agents. This approach guides LLM-based agents to operate within specific parameters, preventing unsafe ...
OpenAI recently launched a new image generation feature in its popular ChatGPT service, utilizing its advanced GPT-4o model. This addition allows users to create images based on text prompts, enhancing the interactive experience within the platform. With access to hundreds of millions of ChatGPT users, OpenAI is at a significant advantage over smaller competitors, who ...
The Model Context Protocol (MCP) has made significant strides with its latest update aimed at enhancing how AI agents communicate with tools and data. OpenAI and Microsoft have both announced their support for MCP, introducing features that improve security and functionality. Key upgrades include a more secure authorization framework, real-time data flow, and efficient request ...
In the past year, interest in voice AI has surged, fueled by advancements in technology that make it simpler for developers to create customized voice agents. Companies like OpenAI are leading the charge with new audio models, while vertical voice agents tailored for specific industries are gaining traction. These agents leverage industry knowledge to enhance ...
Microsoft has introduced two new AI agents, Researcher and Analyst, designed for its Microsoft 365 Copilot platform, set to roll out in April 2025. These tools aim to enhance workplace productivity by integrating internal data with external information to provide deeper insights. The Researcher agent focuses on advanced analysis, helping users create comprehensive strategies and ...
In a recent blog post, the technical co-founder of Benchling shares their transformative experience with OpenAI’s o1 model during Christmas 2025. Initially skeptical, they were amazed at how AI could now tackle complex scientific tasks, prompting a shift in their focus. They dedicated themselves to developing AI assistants specifically for scientists, recognizing that context is ...
In today’s tech update, several exciting developments in AI were announced. OpenAI launched new audio models for voice agents, while Anthropic shared insights on their latest engineering tool. NVIDIA introduced the AgentIQ Toolkit, designed for creating multi-agent systems that boost performance and streamline workflows. Researchers also unveiled an innovative “agentic memory” system for LLM-based agents, ...
OpenAI has launched an API that enables developers to create AI-powered voice agents using new text-to-speech and speech-to-text audio models. These models offer enhanced customization and “steerability,” allowing developers to control how the AI speaks, mimicking styles like a sympathetic customer service agent. The API promises better performance and cost-effectiveness compared to previous models. Additionally, ...
The emergence of AI agents, particularly OpenAI’s Operator, is changing how we work by automating tasks like booking trips and filling forms. However, this innovation also poses new cybersecurity risks, as cybercriminals could use these tools for more sophisticated attacks. Unlike earlier AI models, which only helped with simple tasks, AI agents can now perform ...
