Meta Muse for small business: what it connects to, and what the share link exposure teaches

Meta expanded its Muse agent to small businesses on September 29, 2026, with access to business accounts and tools like QuickBooks, Shopify, and Slack, according to TechTarget. By October 20, Meta had paused Muse’s sharing feature after researchers found over 6,800 shared conversations discoverable through public DNS data, including sensitive business information, according to Ars Technica and The Verge.

This article explains what Muse offers for small businesses and what the share link exposure teaches about AI agent security.

What happened with the share links?

Security researchers from Claroc found that Muse’s share links were easy to discover. When users shared a conversation, the conversation title appeared in the URL subdomain. That meant anyone searching public DNS records could find links without having them shared directly.

According to Ars Technica, the exposed conversations included a marketing company’s internal revenue review marked for its CEO, a small business owner’s loan application, and a workplace staff vacation calendar listing dozens of employees. The Verge confirmed that Meta paused the sharing feature while reviewing the report.

The technical cause was a design decision: putting the conversation title in the subdomain rather than in the URL path made the links discoverable through passive DNS searches.

What does Muse offer for small businesses?

Meta is positioning Muse as a tool for handing off multi-step tasks, not just asking questions. According to TechTarget, business owners can now connect Muse to Facebook and Instagram business accounts, Meta advertising accounts, and third-party tools like QuickBooks, Shopify, and Slack.

The expansion allows users to give Muse broader goals, such as managing a business or finding new customers, rather than simply answering individual questions.

TechTarget reports that Muse reached 5 million downloads in North America in its first three weeks and reached the top of the US App Store charts.

What do analysts say about agent governance?

Wayne Hankins, an analyst at Gartner quoted in TechTarget, said agent security is “ultimately a governance challenge.” His advice: establish clear boundaries around the data, tools, systems, and decisions an agent can access before deployment.

Hankins highlighted human approval for high-risk actions as an important safeguard. “As agent autonomy increases, organizations should implement human approval workflows for sensitive actions,” he said.

He also emphasized logging and monitoring: “Comprehensive logging, observability, audit trails and runtime monitoring are also needed to understand what agents are doing and ensure their actions remain within approved boundaries.”

What this means for your business

Shared links, chat history, and connected accounts are now part of your data footprint when you use AI agents. A bookkeeper sharing a client summary, a retailer sharing a supplier negotiation, or a manager sharing a hiring discussion could expose information beyond the intended audience.

The Muse exposure did not involve a traditional data breach. Users created the share links themselves. The problem was that those links were discoverable by people who were never intended to receive them.

This pattern applies to any AI tool with sharing features. If the sharing mechanism is not carefully designed, information can leak through metadata, URLs, or indexing.

A short policy owners can adopt this week

1. No sensitive conversations in shareable links. If you would not email it to a stranger, do not create a share link for it.

2. Review what is connected. Audit which business accounts and tools are linked to AI agents. Each connection expands the data the agent can access.

3. Assign responsibility. Decide who in your organization can connect business accounts to AI tools. One person connecting their personal curiosity can expose company data.

Frequently asked questions

Is Muse safe to connect to QuickBooks?

That depends on how you configure it. The question is not whether the connection itself is secure, but what data the agent can access and what actions it can take. Review the permissions, limit access to what is necessary, and require approval for sensitive operations.

How do I find links my team has shared?

Most AI tools do not provide a centralized view of shared content. Check with each team member and establish a policy going forward. Some platforms offer admin controls that can restrict or audit sharing.

Should I stop using AI tools that have had security issues?

Not necessarily. Every technology has vulnerabilities discovered over time. The question is how the company responds. Meta paused the feature and is reviewing it. What matters for your business is whether you have controls in place that limit exposure regardless of the platform.

Map your data footprint before connecting

The Agent Opportunity Audit starts by mapping which data an agent would touch before anything is connected. That includes identifying sensitive information, reviewing access requirements, and designing controls. If you are considering connecting AI tools to business systems, the Audit shows what to secure first.