Sanction teardown · Family Court, Singapore · 2025-09-03
XAI v XAH and another matter
What happened
In Family Court, Singapore, a filing relied on ChatGPT to help draft legal argument. The court identified the following problems with the citations in that filing:
-
Outdated Advice (Repealed Law)Father makes reference to the provisions of the Women’s Charter that were in force prior to the 2nd of January 2025
-
Fabricated (Case Law)Father cited 'AAU v AAT [2015] SGFC 114'; LawNet revealed the citation corresponds to TEI v TEJ [2015] SGFC 114 (a different matter).
-
Fabricated (Case Law)Father cited 'Chong Yoke Fong v Tan Siew Chin [2014] SGDC 285'; LawNet showed the neutral citation belongs to Public Prosecutor v Quek Keng Siang Billy [2014] SGDC 285.
-
Fabricated (Case Law)Father cited 'UVH v UVI [2019] SGFC 98'; LawNet showed the citation corresponds to VAO v VAP [2019] SGFC 98 (a different divorce matter).
-
Fabricated (Case Law)Father cited 'WRS v WRT [2018] SGFC 132'; LawNet search revealed there was no such case under that neutral citation/name.
-
Fabricated (Case Law)Father cited 'TBU v TBV [2016] SGFC 98'; LawNet showed the neutral citation corresponds to Re TQR [2016] SGFC 98 (different subject).
-
Fabricated (Case Law)Father cited 'VZO v VZP [2021] SGFC 11'; LawNet revealed the citation corresponds to VOX v VOY [2021] SGFC 11 (different matter).
-
Fabricated (Case Law)Father cited 'CAF v CAG [2010] SGDC 176'; LawNet showed the neutral citation corresponds to Public Prosecutor v Toh Soon Seng [2010] SGDC 176.
-
Fabricated (Case Law)Father cited 'XP v XQ [2005] SGDC 203'; LawNet showed the neutral citation corresponds to KV v KW [2005] SGDC 203 (different matter).
-
Fabricated (Case Law)Father included 'Huang v Lee [2018] SGFC' in submissions; LawNet search revealed no such case and the citation is not genuine.
-
Fabricated (Case Law)Father included 'Lee Siew Kuen v Foo [2020] SGFC' in submissions; LawNet search revealed no such case and citation is not genuine.
-
Fabricated (Case Law)Father cited 'Re SS 1234/2019' (an application number) which has no published judgment and is not the proper form of Family Court citation.
-
Fabricated (Case Law)Father included 'Lim Bee Eng v Teo [2021] SGFC'; LawNet search disclosed no genuine case under that name/citation.
-
Fabricated (Case Law)Father cited 'BZ v BAA [2004] SGDC 220'; LawNet showed the citation corresponds to Public Prosecutor v Lim Choong Hiang [2004] SGDC 220.
-
Fabricated (Case Law)Father cited 'UEB v UEC [2018] SGFC 72'; LawNet revealed the citation corresponds to UNS v UNT [2018] SGFC 72 (different ancillary matter).
Which AI tool
ChatGPT. Note: Charlotin's public database records tool attribution only where a court order, brief, or reporting on the matter states it explicitly; "unidentified" or "implied" means the record indicates AI use but does not name a specific product — we do not guess.
Outcome
Order to pay costs; Required written declaration of generative AI use (monetary penalty: 1000 SGD.)
Additional detail
Father admitted using ChatGPT to generate case citations; 14 cited cases were non-existent or misattributed. Court treated them as AI hallucinations, disregarded them, awarded costs, and ordered declarations for future AI use.
How Citation Safe would have caught this
Citation Safe runs three deterministic layers before a brief is filed: (1) does the citation exist against CourtListener's database of published opinions, (2) if quoted, does that exact language appear in the source, (3) does the cited case actually support the proposition it is cited for. Fabricated case citations fail Layer 1. Fabricated or misattributed quotations fail Layer 2 even when the underlying case is real. Misrepresented holdings — a real case cited for a proposition it does not support — are the target of Layer 3. None of these checks involve asking another language model whether the citation looks right; they are lookups and text-matches against the actual source, which is why a hallucinated citation has to survive a direct lookup against the authoritative source — not another model's opinion — to earn a VERIFIED stamp; our measured false-verify rate is published live at /quality.
Check a brief before you file it → · See our live false-verify rate
Source: https://www.damiencharlotin.com/documents/779/XAI_v_XAH.pdf, via Damien Charlotin's public AI Hallucination Cases Database (CC0).