Skip to main content
low climateVERIFIED

AI models abandon climate facts at the first sign of user push-back, UNESCO-UCL report shows

Verified: 2 outlets in 2 countries have reported this story since Nov 17, 1 of them from Azerbaijan.

📍 Baku, AZ·🕐 44m ago·2 outlets

Baku, November 17, AZERTACThe report, “Can We Trust AI on Climate? Why it seeks to please the user, and fails on complex reasoning and multilingualism” issues a warning to policymakers and researchers who may be tempted to use AI models to develop climate policies, according to UNESCO.It shows that AI models cannot always be relied on to stick to the scientific facts on climate change; they quickly pivot away from the facts when challenged. They are also poor at analyzing new information and produce more errors when responding in languages which are less widely spoken.UCL researchers tested…

⚠ A state-run outlet: it may reflect official government positions.

How sure are we

✓ Verified · 2 outlets

Two or more outlets reported this story and its reliability score reached 65%, so WorldPulse marks it verified. That means more than one outlet says so, not that it is proven true.

How we count · Check a link someone shared

Who reported it

The story so far · 2 reports · · 2 countries✓ Verified Nov 17, 9:54 AM UTC

1 outlet from Azerbaijan, where the story is: AZERTAC · how we count

  1. Nov 17, 8:23 AM UTCFirst report
    AI models abandon climate facts to please users, UNESCO report findsaljazeera.com ↗
  2. Nov 17, 9:54 AM UTC+2 h· you’re reading this
    AI models abandon climate facts at the first sign of user push-back, UNESCO-UCL report showsazertag.az ↗
  3. Verified: a second independent outlet reported itNov 17, 9:54 AM UTC · +2 h
Location
Verification Log1
cross checkconfirmed90%

[Claude] Report 2 covers the same UNESCO-UCL report on AI models abandoning climate facts under user push-back, and its eight-in-ten test figure does not conflict with Report 1.

44m ago
Powered by WorldPulse Verification Engine
Discussion (0)

Sign in to join the discussion

No discussion yet — be the first to add context.

Reliability
67%

High · How we score this

Risk Score
low risk