Notice: This article was created with AI.
What’s It About?
According to the report by Heise, the voice mode of ChatGPT is developing from a mere conversation partner into an assistant that actively takes action. Anyone speaking with ChatGPT Voice can accordingly trigger and manage tasks in connected services such as email, calendar and Slack. For security-relevant actions, confirmation on the screen is provided for.
The source names OpenAI as the originator of the extended voice functions. The voice mode is also meant to be available in the work mode of ChatGPT and thus to be suitable for more complex tasks.
Background & Context
According to the account, voice can be used to start new tasks, query the status of running tasks and give further instructions. Coordinating several tasks at the same time is also meant to be possible: users can switch between different agents without interrupting the conversation. Listening and speaking are meant to run more smoothly, so that the dialogue feels more natural.
Users can also make the current screen content available to ChatGPT as context. The fields of use named range from scheduling appointments through creating documents to interacting with smart devices. According to the source, the functions concern both the desktop app for macOS and Windows and mobile users. They are available only in certain paid plans; exactly which these are and whether that has changed in the meantime is something interested users should check directly with OpenAI.
The confirmation on the screen is a security mechanism: when voice commands have access to emails or calendars, sensitive actions are not meant to be carried out without human oversight. How strictly this check applies in an individual case does not emerge from the available material.
What Does This Mean?
- Voice assistants are changing from pure information systems into tools that trigger tasks in connected services.
- The confirmation on the screen is meant to prevent security-relevant actions from being carried out unnoticed. Users should nonetheless check carefully before confirming.
- Anyone wanting to steer several tasks in parallel could benefit from switching between agents during an ongoing conversation.
- Companies and private users should clarify before deployment which services they connect and which data ChatGPT can see as a result.
Sources
ChatGPT Voice wird agentischer und greift auf externe Dienste zu (heise online)
ChatGPT Voice: Agenten lassen sich mittels gesprochener Sprache steuern (ComputerBase)
Bericht zu ChatGPT Voice mit Agentenfunktionen bei AIbase (AIbase)
This article was created with AI and is based on the listed sources as well as the language model’s training data.
Further Reading: OpenClaw: How a message becomes an acting AI agent