Anthropic is merging the features of Claude Cowork and the standard chat into a single platform, adding new tools for working with documents and presentations. The update is being rolled out gradually and is available only to paid subscribers.
Users have noticed that public chats with the Claude AI are now appearing in Google search results, raising concerns about privacy. This incident highlights the importance of clearly distinguishing between general access and public searchability when using AI platforms.
In a practical test for vulnerability detection in applications, modern AI models demonstrated significant differences in both effectiveness and cost. GPT-5.5 proved to be the most effective, while DeepSeek V4 Pro was the most cost-efficient—an important factor when scaling cybersecurity tools.