I assume if the report was written with newer models it wouldn't have hallucinated so much?

The new AI models make it worst. They make the hallucinations plausible enough to survive reviews.

So you have it so much worst. As analogy would be like you have less clown shoes, and weird fingers, but more forged passports.

Every report about AI problems has this comment. Just one more model, bro.

just one more model. just one more prompt. just one more guardrail. just one more. just. one. more.

It's the 2026 version of "two more weeks"

that always seems to be the case though

they'd just be better, harder to spot hallucinations on a larger scale.