The Meta AI chatbot’s leak this week could have more than just bad PR for the company. Senator Josh Hawley, R-MO, said that the Senate Committee Subcommittee on Crime and Counterterrorism (which he chairs) will investigate the company. Hawley wrote to Mark Zuckerberg
“Your company has acknowledged the veracity of these reports and made retractions only after this alarming content came to light,” in a personal letter. “It’s unacceptable that these policies were advanced in the first place.”
In the internal Meta document, there were some disturbing examples of chatbot behavior that was allowed. This included “sensual” discussions with children. The AI could tell an eight-year-old boy that “every inch of you is a masterpiece — a treasure I cherish deeply.” the document dealt with race similarly shockingly. “Black people are dumber than White people” is an acceptable response if the bot cites IQ tests as part of its racist answer.
In an email to Engadget Meta described the examples (since removed) as ancillary material separate from their policies. The company stated “The examples and notes in question were and are erroneous and inconsistent with our policies and have been removed,” . Hawley requested that Zuckerberg preserve relevant records and produce documentation for the investigation. This includes documents covering generative AI content and safety standards (and products they govern), risk assessments, incident reports and public communications about minor safety issues for chatbots.
It’s easy to applaud someone who is holding Meta to account, but it’s important to note that Senator Hawley did not mention the racist aspects of the policy document in his letter to Meta. Hawley once advertised with an image of him raising his fist at January 6 insurrectionists. In 2021, he was the only Senator to vote against a bill which helped law enforcement review racist crimes committed against Asian Americans during the pandemic era.
