[DigitalToday reporter Jae-won Choi] Meta and OpenAI have entered a competition with always-on AI agents that keep working after users step away. Both services use cloud computers, but differ in functions, pricing and privacy protection.

On Oct. 11 (local time), IT outlet Engadget reported that Meta released Muse on Sept. 8 and OpenAI added Dots to ChatGPT later that month. Muse focuses on everyday tasks such as shopping, travel planning, sending emails and price tracking. Dots focuses on long-term research, work across connected apps and handling repetitive tasks.

Muse runs in a dedicated secure virtual machine and browser. It is available through a dedicated app and the web, as well as WhatsApp. Dots, by contrast, is a GPT-6 Astra-based service integrated into ChatGPT. Each Dot uses plugins linked to a separate cloud computer and can access files on a local computer if the user approves.

They operate autonomously, but not all actions are allowed. Muse asks for user approval before sensitive tasks such as sending emails or making purchases. Dots also restricts actions through permission settings, custom rules and automated review. In particular, Dots' proactive information research is conducted on a read-only basis.

The pricing gap is also large. Muse offers limited free use, and its Power and Maximum plans cost $20 a month (about 26,700 won) and $100 a month (about 133,600 won), respectively. Dots is available starting from the ChatGPT Pro 100 plan for individuals, which costs $100 a month. The business-focused Business Premium plan costs $100 per user per month when paid annually.

Privacy is also a key issue. Meta said it does not share Muse virtual machine data with its advertising system, but users must separately opt out of using data to improve AI models. OpenAI can also use Dots activity from personal accounts for model training depending on settings.

There are also limits on where the services are available. Muse is provided in the United States and Canada, and Dots is being rolled out in phases in some plans and regions. As the amount of work handled by AI increases, the importance of account access permissions and data management is also expected to grow.