Founder Securities AI-Generated Report Misdates Policy from 2024 as 2026
Founder Securities published a real estate research report on June 9, 2026 that attributed a 2024 State Council policy to a 2026 meeting date, exhibiting hallucination characteristics typical of large language models. The report was removed from financial databases after the error was exposed, highlighting gaps in the firm's quality control and fact-checking procedures despite AI's increasing use in investment research.
What happened
On June 9, 2026, Founder Securities published a real estate industry report by analysts Chen Li and Wang Song titled 'Central and Local Policy Efforts, Second-hand House Sales Enter Off-season but Remain at Historical High.' The report attributed a State Council policy on real estate to a June 7, 2026 meeting.
The policy text in the report was actually from the State Council's June 7, 2024 meeting, two years earlier. Verification against Chinese government website records confirmed that while the State Council held a meeting on June 5, 2026, it did not address real estate policy; the June 7, 2026 date mentioned in the report does not correspond to any actual meeting or policy announcement.
The error exhibits characteristics of large language model hallucination, where the model failed to correctly anchor timestamps and instead assembled historical text through probabilistic pattern-matching. Financial platforms Wind and iFinD subsequently removed the report from their searchable databases.
Reported impact
Industry experts noted the error damaged institutional credibility and distorted market expectations, as research reports are core information products that directly affect asset pricing efficiency in capital markets.
The AI-generated error became a vector for further information pollution when some social media outlets amplified the misdate to 2026, demonstrating how AI-generated inaccuracies can multiply across the internet.
Index analysis
The incident exposed Founder Securities' lack of fact-checking mechanisms during high-frequency research production. Analysts prioritized text fluency when using AI to generate and extract content while neglecting cross-verification of facts, allowing the model to output unverified network information as fact.
The error reflects broader industry over-reliance on automation while weakening human oversight. Industry compliance professionals stated AI should function only as an assistive tool with core analysis, investment judgments, and valuations remaining under analyst control, with factual claims requiring human verification and audit layers.
Governance controls
Founder Securities had previously faced compliance scrutiny from the Hunan Securities Regulatory Bureau in January 2024 for research report violations including pre-publication leaks, unauthorized private stock recommendations by analysts, and unqualified employees publishing research opinions without approval.
The incident highlights a systemic quality control and compliance audit gap at Founder Securities. The firm expanded its analyst team by 50 percent (33 new analysts since June 2025, reaching 91 total by June 17, 2026) without corresponding improvements in compliance capacity.
Classification
Sources and evidence
Sources not yet published
This record was researched and written by the Index from original reporting. Its sources are still being verified by an editor, and they are listed here once that check is complete. Until then the record carries no reviewer.