Our practices for covering AI product launches have shifted since our earliest coverage, and naming the specific changes matters more than asserting general improvement.
We slowed down our launch-day coverage
We used to prioritize same-day coverage of major launches. We now more often wait briefly to test claims where possible, accepting being slightly later in exchange for more accurate initial coverage.
We started distinguishing announced from available
Early coverage sometimes blurred the line between a capability being announced and actually being available to use. We now flag this distinction explicitly, since the gap between the two has real implications for readers.
We got more consistent about revisiting launch claims
We didn't reliably follow up on whether a launched product's claimed capabilities held up under real use. We now try to do this more systematically for launches significant enough to have warranted initial coverage.
Why this evolution is worth stating
Product launch coverage is where hype and substance are hardest to distinguish in the moment, and we think being explicit about how we've adjusted our practices helps readers calibrate how much to trust what we publish on launch day specifically.