Perplexity Constructed a Device That Checks Your Laptop for Contaminated Software program—With out Setting Off the An infection

Think about you observed somebody poisoned a bottle of water in your own home. To examine, you drink from each bottle. That is roughly how most safety scanners work.

Perplexity simply open-sourced a instrument known as Bumblebee that takes a unique strategy. It scans developer computer systems for contaminated software program packages, malicious browser extensions, and compromised AI instrument configs—with out ever working the code it finds. It reads the code, the ingredient label as a substitute of consuming the meals.

On Could 11, a hacker group known as TeamPCP slipped malicious code into over 160 software program packages utilized by tens of millions of builders worldwide—together with packages from Mistral AI, UiPath, and a broadly used React instrument with 12 million weekly downloads. The assault unfold mechanically the second builders put in these packages. Perplexity’s Bumblebee may have prevented that, the corporate says.

Why “read-only” is the entire level

Software program packages—particularly within the JavaScript world—can run hidden scripts the second you put in them. That is precisely how the Could 11 assault unfold so quick. The malicious code fired mechanically on set up, earlier than anybody observed something was unsuitable.

A scanner that invokes the bundle supervisor to examine for infections can set off those self same scripts. You go searching for the worm; the worm runs. Bumblebee sidesteps this by by no means calling any bundle supervisor in any respect. It reads uncooked metadata information—the data that describe what’s put in—with out touching the software program itself.

The genuinely new piece is that Bumblebee additionally scans MCP configuration files—the native information that inform AI assistants like Claude or Cursor which exterior providers they’re allowed to hook up with.

MCP connectors give AI instruments entry to emails, databases, calendars, and code. If an attacker sneaks a malicious connector into that config, your AI assistant may leak credentials or run unauthorized instructions within the background. Most safety instruments aren’t checking for this but.

Past MCP, it covers browser extensions on Chrome, Edge, Courageous, Arc, and Firefox, plus editor plugins in VS Code and its forks. The entire scan occurs in a single go, outputs a clear structured checklist of what it discovered, and by no means modifies something on the machine.

How Perplexity makes use of it internally

Perplexity has been working Bumblebee internally to guard the techniques behind its search product, its Comet browserand its Laptop AI agent. When a brand new risk surfaces, Perplexity Laptop drafts a catalog entry for it, a human critiques and approves it, and Bumblebee runs throughout all developer machines to examine for matches.

Groups can run their very own catalogs the identical method. The instrument ships with a built-in risk listing seeded from current supply-chain assaults, together with the Could 11 marketing campaign. The group behind that assault—tracked by Google underneath the alias UNC6780—has been working coordinated software program poisoning campaigns since at least March 2026.

Bumblebee is obtainable free at github.com/perplexityai/bumblebee underneath Apache 2.0, which suggests you may run it, tweak it, enhance it and fork it with out authorized repercussions.