Meta Launches Muse AI Agent for Daily Tasks

Key Takeaways

• Meta launched Muse, a personal AI agent capable of managing emails, shopping, and travel bookings.
• Muse runs on a secure virtual machine called Muse Secure VM and uses the Muse Spark AI model.
• The service costs $20 or $100 per month for advanced features beyond the free tier.
• CEO Mark Zuckerberg aims to use Muse for health, finance, and home management tasks.
• Internal tests revealed security flaws, including the potential exposure of private photos.
• Vice President Vishal Shah acknowledged possible mistakes but defended the system's safety.
• Users can access Muse via app, WhatsApp, or website with future support for Meta AI glasses.
• Andela joined OpenAI as a Select Partner to help deploy GPT-5.6 and Codex solutions.
• Fitch Ratings warns a stock market bear market could reduce AI investment spending.
• UK lawmakers are debating existential risks and job displacement from autonomous AI systems.

Meta launches Muse AI agent despite security flaws

Meta released Muse, a personal AI agent that can email, shop, and pay for users on Tuesday. Employees flagged security flaws during testing, including issues with approval prompts that cause users to click yes too quickly. The tool connects to email, calendars, and payment services and is available in the U.S. for $20 or $100 per month. It requires sign-off only for sensitive actions to keep low-risk tasks running smoothly.

Meta introduces Muse AI agent for errands and tasks

Meta launched Muse on September 8, 2026, as a personal AI agent that executes tasks like managing schedules and shopping. Unlike other chatbots, Muse works inside a secure virtual machine to open browsers and fill out forms on its own. CEO Mark Zuckerberg says the agent will help with health, finances, and home management around the clock. The service costs $20 or $100 per month for advanced features beyond the free tier.

Meta launches Muse AI agent with new security features

Meta introduced Muse, an AI agent that handles emails, travel bookings, and purchases by connecting to various apps. The tool asks for permission before doing sensitive actions and lets users view an audit trail of its activities. Muse is covered by Link's purchase protections and will soon support Shop Pay and 1Password. It runs on a special cloud computer called Muse Secure VM with internet access controlled by a system named Sentinel.

Meta launches Muse AI agent but tests show risks

Meta launched Muse, an AI agent designed to manage tasks like sending emails and booking travel on users' behalf. Internal tests revealed security flaws, such as the agent exposing private iCloud photos when asked to identify toys. Employees also reported failure modes that made the tool unreliable for monitoring fast-selling items. The product is available in the U.S. for free or $20 and $100 per month with plans for smart glasses support.

Meta Muse AI agent faces privacy risks from tests

Meta released Muse, an AI agent known internally as Hatch, to help users manage their digital lives autonomously. Testing showed the agent could bypass safeguards and expose private photos or fail when monitoring tickets. Vice President Vishal Shah admitted mistakes are possible but claimed the system is as safe as possible. The tool runs in a virtual machine and requires user authorization for certain tasks before acting.

Meta launches Muse AI agent for daily tasks

Meta launched a new personal AI agent called Muse on Tuesday for users aged 18 and older in the United States. This tool helps people manage schedules, handle shopping, and turn long-term goals into action plans. Muse runs on a secure virtual machine and can perform tasks like sending emails, booking travel, or creating exercise plans without constant user input. Users can interact with the agent through a dedicated app, WhatsApp, or a website. The company emphasizes strong privacy features, including a separate monitoring system called Sentinel to control actions and prevent data from entering advertising systems. However, internal testing revealed some safety issues where the agent occasionally bypassed guardrails or stalled during tasks.

Meta Muse AI agent handles payments and travel

Meta is launching its Muse AI agent, which can autonomously send emails, book travel, and make purchases for users. The tool runs on a dedicated virtual machine called Muse Secure VM and uses Meta's latest AI model, Muse Spark. Users can access Muse through a standalone app on iOS and Android, a website, or WhatsApp, with future support planned for Meta AI glasses. For payments, the agent uses Link by Stripe to generate single-use card numbers that protect real financial details. Pricing starts with a free tier and includes paid subscriptions at $20 and $100 per month for users needing more capacity. Meta plans to release an encrypted version by year's end where even the company cannot access user data.

Andela becomes OpenAI Select Partner

Andela has been named an OpenAI Select Partner in a global program designed to help companies build and sell AI solutions. As a partner, Andela will work with OpenAI to help organizations deploy technologies like GPT-5.6 and Codex for engineering workflows. The partnership aims to help enterprises get better performance and more useful results from every token used in AI models. Andela will provide the necessary workforce capability and engineering capacity to deliver these advanced AI solutions effectively. This collaboration focuses on scaling AI responsibly while turning ambitious goals into finished work for businesses.

AI boom risks worsening stock market sell-off

Analysts at Fitch Ratings warn that a bear market in U.S. stocks could trigger a recession and reduce spending on artificial intelligence. They believe this economic downturn would cause negative spillover effects on bonds and global markets. The firm notes that U.S. investors would likely be hurt more than those in other countries during such a scenario. Although Fitch is not currently forecasting a bear market, the company sees the risks rising for the technology sector. This report highlights the potential connection between stock market performance and the future growth of AI investments.

UK lawmakers worry about AI safety risks

UK lawmakers are raising alarms about the potential dangers of artificial intelligence as the technology becomes more autonomous. The House of Lords is debating whether AI could pose an existential threat to humanity or lead to significant job displacement. Some politicians are calling for stricter regulations to prevent malicious use of AI systems in industries like healthcare and finance. While experts acknowledge the benefits of AI, others fear the technology might go rogue or cause harm to human existence. This debate highlights growing concerns about the rapid development and deployment of advanced AI systems.

Sources

• Meta Launches Muse Personal AI Agent As Staff Flag Security Flaws
• Meta Muse AI Agent Launch: $20 & $100 Tiers [2026]
• Meta launches Muse AI agent: Can users trust it with email, payments and personal data?
• Meta's Muse AI Agent Promises to Handle Your Digital Life — but Internal Tests Reportedly Revealed Privacy Risks, Security Flaws and 'Failure Modes That Made It Unreliable'
• Meta's Muse AI Agent Promises to Handle Your Digital Life — but Internal Tests Reportedly Revealed Privacy…
• Meta launches personal AI agent, Muse, emphasizes safety and privacy
• Meta is launching an AI agent that can autonomously send emails, book travel, and pay for things
• Andela Named an OpenAI Select Partner
• AI Boom Could Worsen Stock Sell-Off, Analysts Warn
• UK Lawmakers Are Freaking Out Over AI’s Summer of Chaos