NPR reported that the Pentagon is building weapons and projects with artificial intelligence capabilities that officials say "don't always closely adhere" to the agency's own guidelines for AI. According to NPR's summary, the findings come from a congressional watchdog report published in October. The story was written by Chiara Eisner.

NPR's headline frames the matter as whether the Pentagon's guidelines for responsible AI are followed in practice. The summary indicates that the watchdog's findings concern work the agency is currently carrying out, but it does not say how many projects were reviewed.

The summary does not identify the specific weapons or projects involved, describe how they differ from the guidelines, or report any response from Pentagon officials. NPR's account, as summarized, also does not say what steps the agency has taken or plans to take in response to the report.

Under NPR's account, the central issue is the gap between written policy and the work carried out under it. The guidelines are described as the agency's own, and the report is described as finding that the related weapons and projects do not always closely follow them.

Because the summary is the only source available for this account, the details above reflect NPR's characterization alone. The summary names officials only in general terms and does not identify them. The complete account appears in NPR's original article.

Sources: NPR: The Pentagon's guidelines for "responsible AI" aren't always followed, report shows.