Greater accountability would come by requiring a “defined complaint system” that handles reports and alerts users to moderation actions within two weeks while providing an appeals process. Companies would have 24 hours to remove any content deemed illegal, although smaller outlets would have more leeway for responding to requests and user complaints.

The bill would limit Section 230’s ability to to protect companies from actions by federal regulators and state attorneys general, and would have the Government Accountability Office examine the possibility of an FTC-run whistleblower program for online platforms.

If all goes well, this would hold sites to task when they’re abusing the law or are clearly aware of users’ illegal activity. A site that thrives on user uploads of pirated material would have a more difficult time avoiding legal action. This wouldn’t weaken encryption (as the EARN IT Act and other initiatives might), however, and it wouldn’t force sites to carry content.

There’s no certainty this bill and its expected House counterpart will survive Congress and become law. Schatz pointed out that this is not only a bipartisan bill, but is sponsored by both the chairman and ranking member of the Senate committee. It might gain traction where single-party bills have fallen short.

The General Services Administration would be tasked with both providing AI expertise to agencies and researching policy. The Office of Management and Budget would need to devise strategies for investing in and using AI for federal data. The Office of Personnel Management, meanwhile, would identify the skills necessary for AI workers and create relevant jobs. You could also expect an advisory board to tackle AI policy “opportunities and challenges,” while executive agencies would create plans that spure the adoption of AI while protecting privacy and civil liberties.

The 2019 edition of the bill isn’t necessarily going to fare much better than last year’s counterpart, which stalled out after it was sent to committee. This new measure does have some important backers, though, including Facebook, Microsoft and the Internet Association (which involves Amazon and Google). If it ever becomes law, it could go some distance toward incorporating AI in government and parsing the volumes of government data that sometimes go underused.