Rep. Summer Lee on Data Privacy Risks of Centralized Government Data and AI

Centralizing Government Data

• 🎯 Musk and DOGE accessed federal department data sets, initially to cut funding, but the goal has shifted to amalgamating data into one master database.
• 📌 This database will reportedly contain sensitive personal information including Social Security numbers, home addresses, credit card numbers, and medical diagnoses.
• ⚠️ Consolidating data into a single location significantly increases the risk of hacking and data theft.
• 🚀 Data is also being moved from secure government databases into private companies for use by opaque algorithms.

Opaque AI Algorithms and Data Quality

• 🧠 Algorithms are described as opaque, with even their creators not fully understanding them.
• 🧩 Poor data quality and coding errors, like the example of 150-year-olds receiving Social Security, can lead to bad inferences that go unchecked.
• 📊 The quality of data is crucial, especially when it is combined from various sources, and issues with how data is presented can lead to significant problems.

Potential Misuse of AI and Data by Trump Administration

• 📈 An AI system could theoretically be used to make decisions on Medicaid eligibility, veteran benefits, and hiring decisions.
• 🚫 Trump and his administration could use this collected data to profile individuals and deny them benefits, with a known inability to be unbiased.
• ⚖️ A provision in the reconciliation bill aims to ban state regulation of AI for 10 years, preventing states from establishing their own guardrails.

Consequences of Improperly Deployed AI

• ⚠️ Michigan had to refund $21 million to residents wrongfully accused of unemployment fraud by an AI system.
• 🏠 A tenant screening system called Safe Rent was accused of violating the Fair Housing Act due to discriminatory AI algorithms.
• 📚 The "Make America Healthy Again" report cited non-existent studies, highlighting issues with AI in government reporting.

Need for a Methodical Approach to AI

• 🔬 Universities like Carnegie Mellon are working on centers for AI standards and innovation to develop a methodical approach to assessing AI benefits and risks.
• ✅ The focus should be on ensuring AI programs are used safely, accurately, and without bias.
• 💡 The Trump administration's approach should be to develop these standards, not to hand over personal data to unvetted tech billionaires.