morality

You judge side effects based on how much you like them.

In a 2003 study, participants were significantly more likely to label a side effect as 'intentional' when it was negative than when it was positive.

Change 'only when' to 'significantly more likely to' to accurately reflect the asymmetry observed in the study.

Your perception of intent depends entirely on your moral framing.

Where this comes from

Study: Joshua Knobe (2003), Analysis
What the check found: The claim overstates the research finding by implying an absolute binary condition ('only when') rather than a probabilistic asymmetry.
Confidence: Solid, with limits

One of these every day, and every one names the study it came from. When a finding is contested, it says so.