Senator Rosen Questions Microsoft President on AI Security and Adversarial Threats

Adversarial AI and National Security

• ⚠️ Senator Rosen expressed concern about adversaries, specifically mentioning the PRC, potentially co-opting AI for data collection and ideological promotion.
• 💡 Brad Smith, Microsoft's President, highlighted the distinction between open-source AI models and applications, stating Microsoft does not allow employees to use the Deepseek app due to data concerns and propaganda potential.
• 🔑 Microsoft analyzed and modified the Deepseek open-source model to remove harmful elements, emphasizing a security-first approach.

AI and Anti-Semitism

• 🎯 The ADL reported that major generative AI models have perpetuated anti-Semitic stereotypes and conspiracy theories.
• 🤝 Industry is collaborating with civil society to develop standard benchmarks for AI evaluation related to anti-semitism and other forms of hate.
• 💬 While users should have freedom, there's a commitment to prevent AI from being used to create damaging or hateful content.

Data Center Security and Hardware Innovation

• 🔒 Senator Rosen emphasized the importance of data center security alongside energy and water usage, referencing her bipartisan Federal Data Center Enhancement Act.
• ⚡ Dr. Sue discussed AMD's focus on creating more energy-efficient chips and ensuring the security of hardware against breaches as high priorities in their development cycle.
• 📈 Significant improvements in chip efficiency have been observed, with continued focus on further advancements.