Typosquatting bet on a typo; slopsquatting bets on your ai assistant. research now shows nearly one…
typosquatting bet on a typo; slopsquatting bets on your ai assistant. research now shows nearly one in five ai-recommended packages doesn't exist, attackers preregister those hallucinated names, and confirmed real-world packages have racked up tens of thousands of downloads.
the supply chain risk is a name the model invented.
Context
Dark Reading of 14 April 2025 reports a study of 576,000+ code samples from 16 models, by researchers at UTSA, Oklahoma and Virginia Tech, finding 21.7% of package names from open-source models were hallucinated and 5.2% for commercial models, with 205,474 unique invented names. An arXiv paper, 2605.17062, replicates it on five frontier models released from October 2025 to March 2026 across 199,845 prompts and finds hallucination rates of 4.62% to 6.10%, calls it an order-of-magnitude compression of the spread and not a retirement of the threat, and finds 53 registrable names, 41 on PyPI and 12 on npm. Snippets describe a Lasso Security proof of concept in which a hallucinated name, huggingface-cli, was registered empty and downloaded, and Aikido noting malicious packages consistent with the pattern while saying it cannot prove what attackers intend.
Nearly one in five is not supported as stated. The 2025 figure of 21.7% is for open-source models only, with 5.2% for commercial ones, and it is a share of package names in generated code samples, so it is not a rate for all recommended packages, and the 2026 re-evaluation of newer models found about 4.6 to 6.1%. The note presents 2025 open-model data as current. The documented real case is a researcher's proof of concept, not an attacker's, and no text read confirms malicious preregistration with tens of thousands of downloads, so that is unverified. The original paper was not read, only the Dark Reading relay, and who coined slopsquatting was not read. Package hallucination is a real, measured phenomenon.
Watch next
- Dated download counts for real hallucinated-name packages.
Sources
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 20 September 2026 at 03:06 IST. Sources are the papers and datasets the note draws on.
View the original post ↗Embed this note
More notes
The air is now being asked to keep its own ledger
the air is now being asked to keep its own ledger: ecmwf’s aifs compo becomes the first ai model to forecast atmospheric composition globally every three hours, cleanair simulates 365 days of pm2.5 over china in ten seconds, and a unified framework maps six pollutants at one kilometer across the whole country. the air now files its own composition report.
read the note →The current is now being asked to draw its own map
the current is now being asked to draw its own map: china’s langya 2.0 predicts six ocean phenomena including internal waves and mesoscale eddies, a deep net called wenhai resolves eddies globally with air sea flux formulas built in, and scripps infers surface currents from the way temperature patterns deform in satellite images. the ocean now files its own circulation report.
read the note →The soil is now being asked to report its own carbon
the soil is now being asked to report its own carbon: a nix color sensor paired with generative data augmentation predicts soil organic carbon without a lab, random forest drives 74 percent of soil health mapping studies, and sentinel 2 tracks five year carbon change across france and italy from 922 samples. the dirt now files its own carbon account.
read the note →