Corrections log
Every correction, on the record.
When Ground Truth gets something wrong, the fix is made on the page, dated, and recorded here — what changed, why, and whether a rating moved. The log exists so you never have to wonder whether an article was quietly edited. It wasn't; if it changes materially, it appears below.
- 27 Jul 2026 — How to read a “99% accurate” diagnostic-AI claim: The 578-woman ultrasound comparison also appears in the company's own press release, which attributes it to an unpublished sub-analysis held as “data on file.” Our finding that it cannot be traced to any published study is unchanged.
- 27 Jul 2026 — The NHS cancer blood test was “99% accurate.” Which 99%?: We wrote that no public record stated which model version is the UKCA-marked product. PinPoint publishes UK Declarations of Conformity on its own site; the gynaecological one (REC-777, 13 May 2026) names PinPoint_Gynae v1.2. The article and open question 3 have been rewritten; the finding that the MHRA register carries no version field is unchanged.
- 27 Jul 2026 — The NHS cancer blood test was “99% accurate.” Which 99%?: We wrote that the 578-woman ultrasound comparison appears on one page and is attributed to no source. It also appears in PinPoint's own press release of 8 July 2026, which attributes it to “PinPoint sub-analysis, data on file.” Our finding that no published study, dataset or statistical report stands behind it is unchanged and now confirmed by the company's own footnote.
- 27 Jul 2026 — The NHS cancer blood test was “99% accurate.” Which 99%?: We wrote that the 90,000 referral denominator cites nothing. The same press release footnotes it — to hysteroscopy volumes scaled for population growth and to NHS Scotland guidance, not to any count of post-menopausal-bleeding referrals. The chain step now says so and is tagged a proxy figure rather than unsourced.
- 27 Jul 2026 — Google made Gemma “medical.” We gave it the job.: A rating line read “on emergency recognition it is no help at all.” That holds at 27B (MedGemma 8% vs base 8%) but not at 4B, where the medical model flags 24% against its base's 0%. The line now states what holds at both sizes: it adds nothing at 27B and never approaches a newer general model. The band is unchanged.
Spotted an error we haven't? corrections@groundtruth.health — challenges are read by a human and answered against the primary source. The full policy is in the editorial standard.