A report that informed Australia's decision to implement a social media ban for individuals under 16 years old is facing scrutiny after a Guardian analysis indicated it includes citations that appear to be AI-generated. The report, produced by the UK-based Age Check Certification Scheme (ACCS) as part of a A$3.48 million age assurance technology trial, was cited by Communications Minister Anika Wells as evidence of "many effective options" for age verification, paving the way for the ban enacted in December of last year.
The core of the ACCS report details tests of existing age-verification technologies. However, a specific chapter on "emerging technologies" references several journal papers concerning new age-checking methods. An ongoing Senate inquiry, established to evaluate further strengthening the social media ban, received a submission highlighting at least two citation errors within this section. The submission suggested these citations were "AI hallucinated, rather than being based on real sources."
When initially questioned by Guardian Australia, a spokesperson for ACCS denied the use of AI in the report's creation. The organization later acknowledged that ChatGPT had been employed to refine and shorten paragraphs. ACCS maintained that AI was not used to generate the report's content or its cited materials, stating that each link and report was manually verified for authenticity and relevance.
Despite ACCS's assertions, Guardian Australia's own examination of the report identified six problematic references within the "emerging technologies" chapter. These issues included Digital Object Identifiers (DOIs) that either linked to non-existent papers, pointed to incorrect research, or featured mismatched author, journal, and publication year details. In some instances, cited papers existed but did not support the claims attributed to them in the report.
Further attempts by ACCS to correct the flagged citations introduced new inaccuracies. One correction provided a paper that ACCS claimed to have accessed in March 2025, several months before its actual publication date in June 2025. The lead author of that paper also confirmed that the individual named by ACCS had not contributed to the research, and the summary of their work in the report was incorrect.
Professor Christian Downie from the Australian National University, who has previously reviewed government submissions containing AI hallucinations, commented that false citations in official reports could "erode public confidence and trust" in governmental institutions. Independent Senator Fatima Payman criticized the report's "slack referencing" and called for an end to such practices in government-commissioned documents.
The Department of Communications informed the Senate inquiry that it had met with ACCS, which attributed the faulty links to issues arising after the report's publication, rather than fabrication. The department noted the difficulty in retrospectively verifying this explanation.
The controversy follows the implementation of Australia's under-16 social media ban in December of last year. Initial data from the eSafety Commissioner's three-month review indicated a modest reduction in social media use among children, with a significant number still accessing platforms. The eSafety Commissioner, Julie Inman Grant, noted that the primary reason for this limited reduction appeared to be "ineffective implementation of age assurance measures by platforms." The ongoing Senate inquiry continues to examine the effectiveness and enforcement of the ban, with social media companies urging caution regarding early data.