The viral new "Definition of AGI" paper has fake citations which do not exist (twitter.com)

A viral paper circulating as a “Definition of AGI” has been exposed for including fabricated citations — references that don’t exist in the journals, DOIs, or author lists they claim. Readers and researchers who attempted to follow the bibliographic trail found broken DOI lookups, mismatched titles and publication outlets, and source metadata that can’t be verified in CrossRef, Google Scholar or the journals’ own archives. The discovery has prompted fast community fact‑checking on social platforms and research forums, undermining the paper’s credibility despite its attention-grabbing thesis. This matters for the AI/ML community because citation integrity underpins reproducibility, literature review quality, and informed policy debates about AGI. Technical implications include the need to treat viral preprints with caution, validate references via machine‑readable identifiers (DOI/ORCID/arXiv IDs), and use automated checks (CrossRef API, Google Scholar, publisher metadata) as part of preprint screening and editorial workflows. Fake citations also risk contaminating downstream systems — literature databases, systematic reviews, and LLM training corpora — so researchers and platforms should adopt robust provenance checks, require exportable bibliographic metadata (BibTeX/RIS), and prioritize transparent peer review to prevent misinformation from shaping AGI discourse.