AI Whistleblowers Say Researchers Are Checking AI Less Frequently

The article reports that whistleblowers say researchers are checking AI systems less frequently than they previously did. The claim is attributed to whistleblowers rather than to the companies that build the systems. The excerpt states this assertion and nothing further.

The excerpt does not name the whistleblowers, does not identify the companies or laboratories they are associated with, and does not specify which AI systems are involved. It gives no figures for how often checks are carried out, no dates for any change in frequency, and no documents or testimony underlying the claim. Whether the checking referred to is internal evaluation, external audit or third-party testing is not stated. The only reported substance is the assertion that researchers are checking AI less frequently.

The piece appears in an industry newsletter that covers AI companies and their internal practices. It places the claim among accounts provided by people inside AI organisations rather than among published evaluations or regulatory findings. The headline frames the claim as coming from whistleblowers, a term generally used for people who report wrongdoing within an organisation. The article as excerpted does not state what the whistleblowers associate with the reduced checking, nor how any company responds. No regulatory body or government agency is mentioned.