Meta's AI chatbot guidelines: US senator to launch investigation

The leak of a 200-page guideline on permitted and prohibited topics for the training and operation of Metas chatbots could have legal consequences in the USA. In addition to a relaxed approach to everyday racism, the guideline contains paragraphs that set the framework for conversations about sexuality with minors. While it prohibits the actual description of sexual acts with minors, the rules leave a lot of leeway when describing romantic encounters and suggestions of sexual interest. Romantic role-playing and flirting were explicitly allowed, according to Reuters. According to the document, it was approved by Meta's legal, policy and technical staff, including the chief ethics officer.

After Reuters reported on the leaked set of rules, Republican Senator Josh Hawley from Missouri announced that he would launch an official investigation. Hawley is head of the Subcommittee on Criminal Justice and Counterterrorism, which is to take over the investigation, according to CNBC. The announcement was made via an X-rated post by the Republican senator, in which Hawley both denounced Big Tech's greed for money and demanded that tech companies keep their hands off children.

The senator also demanded from Meta CEO Mark Zuckerberg that his company should preserve possible evidence, including emails. The investigation should determine whether Meta's chatbots enable exploitation, deception or other criminal acts against children. It is also intended to clarify whether Meta misled the public or the supervisory authorities regarding its safety precautions.

(pst)