US AI evaluation framework partly revealed amid secrecy concerns
us government partially reveals secret ai evaluation framework after foia request, but most details remain redacted. transparency advocates continue to push for disclosure.
the us government has partially revealed its secret ai evaluation framework in response to a freedom of information act (foia) request from protect democracy. the framework, which aims to assess the risks of 'frontier' ai models, remains largely opaque, with most details redacted. the release of 132 pages of records follows a lawsuit filed by protect democracy to obtain the policy. transparency advocates, including the group's lead counsel deana el-mallawany, argue that secrecy in ai regulation can lead to corruption, autocratic abuse, and mistakes. the government has agreed to produce its secret review framework via this litigation.