Claude Code's "don't ask again" button is leaking passwords and API keys to public GitHub repos

Two of the leading players in AI coding are expanding voice capabilities for their coding agents, while another competitor remains silent on the matter. Cursor and Google have both announced features that allow users to engage in verbal conversations with coding agents, contrasting with the predominantly silent approach of Claude Code.

Cursor introduced Voice Chat with Agents, powered by Grok Voice, enabling developers to interact with the agent through spoken commands and receive real-time updates. Similarly, Google unveiled Voice mode in its Antigravity platform, utilizing Gemini Live to facilitate spoken conversations within its desktop app and command-line tool.

Nick Van der Auwermeulen, a developer relations engineer at Google, emphasized the importance of this two-way interaction, drawing a parallel between the fluid communication of human colleagues and the more rigid process of working with coding agents.

Google's voice mode keeps a live call open throughout the coding process, allowing developers to hear updates from the agent without interrupting their workflow. The agent can also be interrupted mid-sentence to provide further instructions or ask questions. When the agent completes its task, it delivers a written summary of the changes made, accompanied by a button to review the code modifications.

In contrast, Anthropic's Claude Code, while offering dictation capabilities, does not provide full two-way voice functionality. Users can only dictate changes and receive textual updates. This disparity highlights the differing approaches among major players in the AI coding space, with Cursor and Google prioritizing interactive, spoken communication, while Anthropic maintains a more traditional, text-based interaction model.