Meta Paid $1.2 Million for Muse AI Agent Bugs, Including $300,000 Pre-Launch Flaw

Meta has paid roughly $1.2 million in bug bounties for its Muse AI agent, including $300,000 for a pre-launch flaw that exposed data on internal test systems. SecurityWeek reported the payouts on November 4, 2026.

Muse is a personal AI agent that acts on a user's behalf across email, calendar, messaging and other accounts, with each user's agent running on its own virtual machine. Meta is the Menlo Park social technology company behind Facebook, Instagram, WhatsApp and Messenger, according to akta.pro company records.

Inside Meta's Muse Bounty Program

Meta's David Singleton said in a blog post that the program ran privately from April and went public at the agent's September 8 launch. Of the more than 630 reports received, 64 were valid. The $300,000 reward was handed out four months before launch, while Muse was at an early internal testing stage, and the researcher demonstrated access to conversations and tokens on pre-release virtual machines used by Meta staff to test the product.

"An authentication check treated a header added by a local HTTPS proxy as proof that a request came from a trusted source. During a system migration, the environment gained a second proxy layer. That meant requests from outside could carry the trusted header too," Singleton explained. Meta took the service offline, fixed the bug and stopped treating the local machine as trusted.

A $90,000 bounty went to a report involving a malicious web page that, when Muse was asked to summarize it, mimicked status messages from the agent's browsing tool. "It got the agent to execute some actions which should have been caught," Singleton noted. Meta added detection signals and changed how Muse treats suspicious tool output, and said its classifiers now catch the attack and 100 variants in testing.

Another $90,000 went to a report showing that specific formatting could slip one-time passcodes past Muse's email redaction filters, which Meta described as one piece of an end-to-end attack. A $50,000 reward covered an Android app flaw that let a malicious app pass a link to a privileged web view. "The link handler recognized a shared-page path without validating its host and loaded it," Singleton wrote. A $1,000 bounty covered a Mac app flaw that let an attacker with access to the computer hijack the auto-update feature to execute code with the app's permissions.

Meta pays up to $300,000 per valid report, which it claims is several times more than the top rewards advertised by other leading AI developers. Prompt injection attacks affecting a single user can earn up to $130,000, while a fleetwide compromise can earn as much as $250,000.

Meta's Scale Behind the Muse Program

Meta operates Facebook, Instagram, WhatsApp and Messenger for over 3.56 billion daily users and produces approximately $215 billion in revenue, mostly from advertising, according to akta.pro company records. The company was founded in 2004 and is led by founder, chairman and chief executive Mark Zuckerberg, with Susan Li as chief financial officer and Andrew Bosworth as chief technology officer and head of Reality Labs.

Earlier Meta Announcements

Meta's Muse bounty program went public at the agent's September 8 launch, after running privately from April. On November 3, 2026, Meta launched Shayek, a Saudi Arabia scam awareness campaign with Gmedia, to help users pause before clicking or sharing, Arabnews reported. That same day, PCMag reported that Meta's third-generation Ray-Ban smart glasses add an extra hour of battery life, a six-mic array and two new frame styles, raising the price from $379 to $449.

What Comes Next for Muse

Meta said its classifiers now catch the browsing-tool attack and 100 variants in testing, and the company has changed how Muse treats tool output that looks suspicious.