Pluto Security raises $42M: vibe-coded apps are 60% of AI risk

Pluto Security came out of stealth with $42 million in funding to watch the AI tools, agents and apps that employees bring to work. The number that matters is not the round. CEO Shahar Bahat told TechCrunch that about 60% of the risk Pluto finds across its customers sits in vibe-coded tools built by non-developers. Finance departments are the largest source. Developers rank fourth.

What actually happened

Per TechCrunch, the $42 million is two rounds: an $8.5 million seed led by TLV Partners, and a $34 million Series A led by Jump Capital. Pluto started to sell in April. It reports more than $2 million in annual recurring revenue and more than 30 enterprise customers, including Oscar Health, Workato, Sixt and AlphaSense.

The product monitors AI tools, skills, extensions, MCP servers, and the apps and agents that employees make. It connects to endpoint and device-management systems, and it can stop an agent mid-session, for example when the agent asks for more permissions than the user intended.

Three findings from the article stand out:

• After finance, the next sources of risk are marketing and legal, then developers.
• Bahat says a new AI tool can show up inside a company within about two hours of its release.
• Pluto says it found malicious activity in every proof-of-concept. Some came from extension packages or skills.

One example: a finance employee at a Fortune 500 company vibe-coded an app to manage customer and churn data. The app was open to the internet.

Why vibe-coded apps matter for your business

Your riskiest developer may be your bookkeeper. Small companies do not have a security team that watches for this. The person who built a churn tracker in an afternoon wanted to do a better job, not to leak a customer list. But a login screen that no one tested and a database key in the browser are the same leak, whoever wrote the code. We wrote about exposed Supabase keys in vibe-coded apps earlier.

Do the inventory before you buy the tool. You do not need a $42 million platform to start. Ask each team one question: "What did you build or connect with AI this year?" List every app, every MCP connector, every browser extension, and where each one stores data.

Then pick: harden it or replace it. Some of those tools are worth keeping. Give them real authentication, take them off the public internet, and move the secrets server-side. The rest go.

Sources: TechCrunch.