Australia's federal parliamentary committees have a contamination problem. Guardian Australia found that AI-generated submissions are flooding inquiries with fabricated academic citations, nonexistent studies attributed to real researchers, and hallucinated court references — and some of this invented material has already been cited in official committee reports. The system designed to channel public input into policymaking is now a vector for synthetic misinformation at scale. The scale of the vulnerability is structural, not incidental. Parliamentary inquiries are legally obligated to consider all submissions. Committees dealing with hundreds or thousands of submissions lack the resources to verify every footnote. Liberal Senator Paul Scarr, who sat on a committee that received a submission containing fake allegations against KPMG in 2023, warned that decision-makers are now making policy on the basis of information that may not exist. His remedy — "go the extra mile in terms of checking footnotes" — is a manual patch on an industrial-scale problem. Labor MP Louise Miller-Frost, chairing the social policy and legal affairs committee, offered a more nuanced but equally inadequate response. She argued that AI-generated submissions should not be discarded outright, noting that AI tools help individuals who lack resources to participate in the democratic process. This is a real benefit. But the tradeoff is stark: broadening access also broadens the attack surface for misinformation, and no one is building the filter. The institutional response amounts to a disclaimer. Senate guidance now warns submitters that AI use "can present risks to the quality of information" and places accuracy responsibility on the individual. Senate President Sue Lines emphasized that participants who discover their evidence is misleading have "an obligation to correct it." This is a liability transfer dressed as governance — the institution accepts the material, uses it in reports, but assigns fault to the submitter after the damage is done. The Deloitte incident is the canary. Last year the consultancy issued a partial refund to the federal government after AI tools embedded fake references and a fabricated court citation in a $440,000 report. If a Big Four firm with professional quality controls cannot prevent AI hallucinations from reaching official output, expecting individual citizens or small organizations to do so is fantasy. What is missing is any systemic verification infrastructure. No automated citation-checking. No flagging mechanism for submissions generated primarily by AI. No audit trail distinguishing AI-assisted from AI-generated content. The committees are flying blind, relying on the statistical hope that hallucinated submissions are a "minority" that gets "evened out" by legitimate ones — Scarr's own words. That is not quality control. That is prayer. The twenty-year trajectory here is corrosive. As AI tools become cheaper and more fluent, the ratio of synthetic to genuine submissions will shift relentlessly. Without structural intervention, parliamentary inquiries will gradually lose epistemic integrity — the very quality that makes them worth conducting. The institution risks becoming a laundering mechanism for AI-generated policy narratives, with real democratic participation crowded out by volume.
