SINGAPORE – Starting October 28, users in Singapore can access Meta's Muse AI through the Apple App Store or Google Play Store. This marks Muse as a pioneer in offering an artificial intelligence agent tailored for mainstream consumers.
Muse's introduction here follows its initial launch in the United States and Canada on September 8, with a subsequent rollout in Japan on October 23. It achieved a remarkable milestone, surpassing 10 million downloads in just 40 days, and quickly became the most downloaded free app on the Apple App Store in Singapore.
Capabilities of Muse AI Compared to Other Tools
To maximize its functionality, users will need to provide Muse access to their Gmail, Instagram, and Calendar accounts. This enables the AI to handle tasks like sending reminders or managing appointments. Muse can also integrate with other popular AI tools, including OpenAI's ChatGPT, Anthropic's Claude, and Google's Gemini, linking them to applications like Gmail and Google Calendar.
A recent evaluation highlighted the varying capabilities of these AI tools. For basic functions, such as drafting emails, Muse, along with the other tools, performed well. However, Muse stood out when it came to retail tasks. It not only suggested colored contact lenses available in Singapore but also accurately provided real-time stock levels and facilitated purchases.
When tasked with organizing a night out, all four AIs offered useful recommendations regarding schedules and locations; however, Muse took the initiative further. It sourced live showtimes, checked seat availability, and calculated total costs for different ticket options. In comparison, Gemini occasionally struggled with accuracy, providing wrong information on showtimes and availability.
Muse is available for free, albeit with a weekly usage limit deemed sufficient for typical user needs. For those requiring enhanced functionalities, subscription plans are available, priced between $22 and $139, offering more tokens or credits for a broader range of activities.
Potential Risks of Using AI Applications
User connections with Muse will expand to include various apps like those from Singapore Airlines, Shopee, Lazada, Grab, and more. This integration allows the AI to find the best deals on tickets and items, manage grocery orders for meals, and suggest travel experiences tailored to individual preferences.
To facilitate transactions, users must store payment card details in Link by Stripe, a third-party digital wallet that generates unique one-time card numbers for each purchase, ensuring their actual card details remain secure.
Yifan Jia, founder of AIDX Tech, highlighted the risk of prompt injection, where attackers embed malicious instructions within content that the AI reads, potentially leading to unauthorized actions.
This includes scenarios where an AI agent inadvertently forwards confidential documents due to hidden commands embedded in seemingly benign content. Users need to remain vigilant as they may start relying on AI too heavily, which could lead to missed automated actions and decisions.
Best Practices for Safe AI Use
When utilizing AI agents like Muse, it's critical to limit permissions according to the task at hand. April Chin, co-CEO of Resaro, suggests that if an agent manages calendars, it may not need access to an email inbox.
- Connect only one app at a time.
- Withhold sensitive data, such as banking information, until clear on what the agent can and cannot do.
- Opt for read-only access to external applications when possible, to prevent unwanted changes or purchases.
Regularly reviewing the AI’s activity log and restricting access when tasks are no longer needed is also advisable. Always require human approval for irreversible actions, such as making payments or sending messages.
Considerations for Professional Use
Connecting personal AI agents to work-related emails or systems should be avoided unless management explicitly permits it, as this poses risks to corporate information security.
A compromised AI linked to workplace accounts could provide hackers a direct pathway to private corporate systems. Employers should enforce clear policies regarding AI use, implement access controls, and educate employees on approved AI tools to manage potential risks effectively.
