Vulnerability / research
GhostSquatting exposes abandoned names in AI instructions
Reported
What happened
Researchers registered abandoned bucket and package names referenced by AI instructions, then observed production requests and installations of placeholder packages.
Parties and reporting
AI tools and systems: Agent instruction files, Amazon S3, Google Cloud Storage, npm, PyPI
Outcome
Researchers notified affected organizations and retained claimed names for return; the work demonstrated exposure rather than a malicious theft campaign.
Operating lesson
Audit resource ownership in agent instructions and pin dependencies to verified, maintained sources.
Evidence and limits
Some observed upload requests were rejected by AWS. Potential model theft or code execution must not be presented as confirmed malicious exploitation.
Sources
- GhostSquatting research
Primary source ·