The Government's New AI Chatbot Is a Dumpster Fire of Transparency (And Nobody's Talking About It)

Remember when we thought the biggest threat to democracy was social media? Those were simpler times.

Today, Blake Crawford and I published an open letter to Congress regarding something far more insidious: the Department of Government Efficiency's new AI chatbot, GSAi.

According to a recent Wired article, this system has been quietly deployed to 1,500 federal workers. What hasn't been shared? Literally everything that matters:

• Security protocols? Mysterious.
• Training data? Your guess is good as mine.
• Oversight mechanisms? Apparently optional.
• Ethical framework? Let me know if you find it.
  

This is the digital equivalent of handing out loaded weapons at a kindergarten and calling it "self-defense training." It's not just negligent—it's willfully obtuse.

As researchers who have spent years documenting how AI systems amplify existing biases, we're not just concerned—we're sounding every alarm we can reach. The same week Bank of America announced its data breach affecting half a million people, our government decided to play fast and loose with a technology notorious for security vulnerabilities and algorithmic bias.

Our letter outlines six non-negotiable requirements for ethical AI implementation in government. But really, they boil down to this: don't build AI systems that make terrible human decisions faster and at greater scale.

Those billions-deep in the AI race will happily tell you that regulation just "slows innovation." And I'm here to tell you that's exactly the point. Some things—like democracy, equity, and basic human rights—are worth the pause.

Read our full letter and join us in demanding accountability. Because efficiency without ethics isn't progress—it's just automating our mistakes with better PR.