David Singleton said 64 of more than 630 reports were valid, led by an authentication bug that exposed internal testers' tokens and conversations.
By [Ryan Merket](https://runtimewire.com/author/ryan-merket)
· Published
Primary source: [muse.ai](https://muse.ai/blog/securing-muse)
Why it matters #
Muse asks users to hand an agent access to email, accounts and purchases, so its security failures can become account failures. Meta's $1.2M payout and a $130,000 ceiling for a prompt injection against one user show how the company prices agent-specific attacks.
Meta's Muse personal AI agent has paid about $1.2M in bug bounties, including $300,000 for an authentication flaw found four months before the product launched, David Singleton (@dps) said in a thread on X. The program received more than 630 submissions and produced 64 valid findings, according to Singleton's accompanying blog post on securing Muse.…
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.