David Sacks' two-track vision for AI is now in full view. In his role as White House AI czar, he spent over a year defending the technology from federal oversight, arguing that regulation would cede the economic century to China. That changed in April, when Anthropic revealed Mythos, an AI model so adept at finding software vulnerabilities that Microsoft and JP Morgan's Jamie Dimon called the Oval Office.
Sacks was outmaneuvered internally. White House Chief of Staff Susie Wiles and Treasury Secretary Scott Bessent viewed total deregulation as a political liability. Their initial proposal for a 90-day government review of new models triggered a revolt from Silicon Valley executives like Mark Zuckerberg and Marc Andreessen, who called Trump directly. The president briefly canceled the signing ceremony.
"The 'beautiful baby' era of AI is over."
- The Daily
Wiles and Bessent spent two weeks salvaging a deal. The final executive order, signed quietly on a Tuesday, mandates only a voluntary 30-day review and explicitly prohibits government pre-clearance. It was a tactical retreat that allowed the administration to claim action without alienating its donor base. The order signals a door to regulation is no longer locked, but the industry is banking on no crisis emerging before the election.
Parallel to his regulatory pivot, Sacks is pushing a new commercial agenda focused on AI's immediate, pragmatic use. He is now focused on AI software that solves manual workflows at large financial institutions, a shift from libertarian theorizing to utility. On The a16z Show, he highlighted the deployment of multilingual voice agents for debt collections and AI tools in fixed-income trading at major banks.
His urgency is driven by a stark reality: the most successful application of AI in finance today is fraud. Plaid CEO Zach Perret notes financial crime is growing nearly 20% annually because AI bots now handle the entire scam operation, from building trust to draining accounts. Defense is still manual. The immediate industry priority, which Sacks endorses, is building an anti-fraud network that scores user trustworthiness across apps and banks.
"The biggest current use case for AI in finance is fraudsters committing fraud."
- Zach Perret, The a16z Show
This leaves Sacks navigating between two fires. Populist pressure is mounting from both flanks. Steve Bannon warns of "brolegarchs" and demands a Geneva-style arms treaty for AI, while Bernie Sanders proposes a 50% public ownership stake in firms like OpenAI. The industry's response is a double game - OpenAI now publicly encourages "rigorous rules" while its founders fund hands-off candidates. With a third to over half of U.S. GDP growth now tied to AI, the political calculus is as much about economics as safety.
