Sources checked
Record the claim, measurement, and unknowns
An original newsroom note can record the claim, who measured it, the test setup, and what is not yet known. Separate announced availability from a product you can actually access, and separate API prices from consumer subscriptions.
Keep comparison qualifiers in the published story
A comparison chart may omit unsuccessful tasks or use a different budget for each system. Preserve those qualifications in the article rather than rewriting the vendor’s claim as an independent finding.
What the post is evidence of
It is evidence that the company published those words on that date. If they announce a model, a price, or a partnership, the post supports that they said so.
It does not, by itself, support a benchmark, a customer count, or a safety property unless the post links a method you can read. Follow the link.
If there is no link, say the figure is the company’s claim.
Watch the nouns. “Available” can mean a waitlist.
“Open” can mean a weight file with a restrictive license. “Beats” can mean one metric, on their harness, against a baseline they chose.
Copy the metric name. A headline that drops the metric has already editorialized.
How to write the missing piece
Add one sentence: what would have to be true for the claim to change a decision, and whether the post supplies it. A price change supplies a decision if you pay that price.
A demo video does not supply a decision about your data. You can still report the demo.
You report it as a demo.
What to record before you rely on it
Keep the post URL, the date, and the sentence you are treating as the claim. Next to it, write “company’s measurement” or “not measured here.”
That label is the whole practice. Keep the model id or the paper citation, the date you read it, and the inputs you used beside that note, so a later change can be compared with the same test.
When the result is a failure, store the artifact and change the design. Do not relax the check until the model happens to pass.
Name a person who owns that failure.
Common questions
Should we wait for an independent test before reporting?
No. Report the announcement and label it. Waiting without labeling is how rumors fill the gap.
Pretending the announcement is the independent test is the other failure.
What about a number in a keynote slide?
Treat it as the speaker’s claim. If the slide has no definition, say that the definition was not on the slide.
Do not invent the denominator.
Sources & further reading
How this story was made
Written by Kristian Kostov with AI assistance and checked against the linked sources. Company performance claims are attributed to the company. Analysis reflects AiLookout’s interpretation; we have not independently tested the products discussed. Cover photography is illustrative and does not depict the specific announcement or product.
Our editorial standards





