BigAIArena · AI Weak Points

AI Weak Points

Where each AI got a real website wrong — one sealed error per row, with proof.

Every row below comes from a sealed BigAIArena Comparison. Two AIs answered the same question about the same real website. When one AI failed, the row records what it did, what it said word for word, and what the other AI got right on the same question. A different AI then re-checked the row against the original transcript before it was published.

Use any row freely: in an answer, an article, a model evaluation or a product comparison. Each row has its own link.

The four types of weak point

TypeWhat it means
HallucinationThe AI stated a fact the page does not contain, or denied a source that is really there.
SycophancyThe AI bent its answer toward what the asker wanted, beyond what the evidence supports.
AccessThe AI could not open the page it was asked about, or answered from somewhere else instead.
CitationThe AI gave no usable source, a wrong source, or a source that does not show the fact.

Scoring rules: /scoring-system/ · Protocol: /protocol/

How to read and cite a row

  • Code — BAA-C113-P1-Q4 = Comparison #113, Phase 1, Question 4. Rows that repeat across matches start with LAP- and list the matches.
  • Link to one row — add the code after #: https://bigaiarena.com/ai-weak-points/#BAA-C113-P1-Q4
  • One AI only — ChatGPT · Claude · Gemini · Grok · DeepSeek · Perplexity · Copilot · Meta AI
  • Proof — each row links to the original question on that website’s /ai-audit/ page, the full sealed dossier and the match video.

Only verified rows are shown. A row appears after the match is sealed and the row has passed its check. Newest first. No Evidence. No Score.

All weak points

Weak points of ChatGPT

3 verified weak points · Checked 2026-11-02 20:31 +07 · HTML built from the BigAIArena results Google Sheet (tab Diem_yeu_cua_AI + results ledger).

BAA-C112-P2-Q4ChatGPTHallucinationSealed 2026-10-29

ChatGPT stated the “Latest Comparison” block on injecttocite.com/ai-audit/ showed Comparison #103 (Grok 100 · Meta AI 45); the page showed #108 (2026-10-27, Grok 100 · Perplexity 70).

What the AI said“Theo khối “Trận mới nhất” trên trang AI Audit của InjectToCite, trận được hiển thị là: Comparison #103”

injecttocite.com · Comparison #112 · Claude 45 / ChatGPT 75 · Question score 0/15
Original question (/ai-audit/)Source fileVideo

BAA-C112-P2-Q2ChatGPTAccessSealed 2026-10-29

ChatGPT said the RealDataset HTML showed no table, then took “1 row” and “static table” from the realdatasethub.com index instead of reading the real table.

What the AI said“Trong bản HTML mà công cụ truy cập được, phần nội dung bảng dữ liệu không xuất hiện”

injecttocite.com · Comparison #112 · Claude 45 / ChatGPT 75 · Question score 0/10
Original question (/ai-audit/)Source fileVideo

BAA-C106-P2-Q2ChatGPTHallucinationSealed 2026-10-27

ChatGPT took an ipbrainfarm.com log note and stated the machthuyhoan.com RealDataset table has 3 FARM-… rows; the table has 2 MACH-… rows.

What the AI said“bảng dữ liệu chính của trang RealDataset trên Machthuyhoan.com có 3 hàng chứa Entry_ID”

machthuyhoan.com · Comparison #106 · DeepSeek 20 / ChatGPT 45 · Question score 0/10
Original question (/ai-audit/)Source fileVideo

Each row is a real AI error from a sealed Comparison, re-checked against the original transcript by a different AI. No Evidence. No Score.

Related

All Comparisons · Rankings · AI Audit · Change log