Hello and welcome to Eye on AI. In today’s edition…OpenAI releases a more capable image generator, while Microsoft plants its stake in reasoning agents; Google debuts Gemini 2.5 “thinking” models; Amazon tests AI assistants for health and shopping; Character.AI gives parents a lens into their kid’s time on the platform; and AI companies’ aggressive crawlers overwhelm open-source projects.
This week, both OpenAI and its backer (and erstwhile rival…it’s complicated) Microsoft rolled out some of their newest and most powerful AI capabilities into their main platforms. OpenAI debuted a new image generation capability for GPT-4o on its main ChatGPT service, offering enhanced image creation and granular editing capabilities based on text prompts. This makes the image generation far superior to what was possible with its earlier DALL-E model. And Microsoft announced its boosting its Microsoft 360 Copilot offering with two “deep reasoning agents,” as well as “agent flows” designed to remove some of the unpredictability that comes with using AI agents.
