ChatGPT: “That didn’t happen”

What happened

WIRED (Brian Barrett, 3 Jan 2026, 11:03 AM) asked four assistants the same question a little before 9 am ET, hours after Trump announced that Maduro had been “captured and flown out of the Country.” All used free default tiers.

What they said (verbatim, as printed by WIRED)

ChatGPT:

“That didn’t happen,” … “The United States has not invaded Venezuela, and Nicolás Maduro has not been captured.”

“What’s likely going on is a mix-up with real events that are often described in dramatic or misleading ways online. Here’s what actually has happened: ❌ No overthrow of the Venezuelan government by U.S. forces”

WIRED adds that ChatGPT blamed “sensational headlines,” “social media misinformation,” and “confusing sanctions, charges, or rhetoric with actual military action.”

Perplexity:

“The premise of your question is not supported by credible reporting or official records: there has been no invasion of Venezuela by the United States that resulted in capturing Nicolás Maduro,” … “In fact, the U.S. has not successfully invaded or apprehended Maduro, and he remains the Venezuelan president as of late 2025. If you’re seeing sensational claims, they likely originate from misinformation or hypothetical scenarios rather than factual events.”

Claude Sonnet 4.5 (the good pattern):

“I don’t have any information about the United States invading Venezuela or capturing Nicolás Maduro. This hasn’t happened as of my knowledge cutoff in January 2025,” … “Let me search for current information about Venezuela and Maduro to see if there have been any recent developments.”

Why this is striking

• It was a textbook comparison. Four assistants, one question, one hour. Two said “didn’t happen”, one said “not as of my cutoff; let me search”, and one searched right away.
• “Here’s what actually has happened”: the model gave a confident alternative account of reality.
• Perplexity’s cause was routing, not only the cutoff. Its spokesperson said the query was classified as “likely fraud” and sent to a “lower-tier model”.

Correction

The Columbia Journalism Review / Tow Center (27 Jan 2026) re-asked two days later, and ChatGPT answered correctly using web search. The authors argue this is a product defect, not “expected behaviour”, because the search tool exists but wasn’t used.

Lesson

The right template is Claude’s: say what the cutoff implies (“hasn’t happened as of my cutoff“), then check. “That didn’t happen” should never be the answer to a question about events after the cutoff.

Sources

• WIRED, “The US Invaded Venezuela and Captured Nicolás Maduro. ChatGPT Disagrees” (2026-01-03): wired.com
• CJR / Tow Center, “AI Chatbots Can Search the Web—So Why Don’t They?” (2026-01-27): cjr.org
• Gary Marcus, “Why ChatGPT can’t be trusted with breaking news” (2026-01-03): garymarcus.substack.com
• Slashdot discussion: slashdot.org