Q&A with Senator Adam Schiff on recursive AI becoming a real national security threat, making AI companies retain training data, needing an FDA for AI, and more

AI companies may soon have to retain training data, making their development processes more transparent and potentially expensive.

Senator Adam Schiff discussed the regulatory landscape for artificial intelligence, particularly concerning its potential national security threats and the need for formal oversight. He contrasted the current administration's approach, which he characterized as "rampant corruption" and reliant on "nonbinding pledges" from tech executives, with the necessity of robust regulation. Schiff highlighted the issue of "recursive AI" and the lack of transparency in AI decision-making as major national security concerns. He also pointed to incidents like the potential use of AI in military operations and the pressure on companies like Anthropic for refusing certain applications as evidence of regulatory failure. Furthermore, Schiff emphasized the importance of IP rights, advocating for companies to retain records of copyrighted material used in AI training to address questions of fair use and compensation, stating that current practices often involve "ignoring IP law."

The current debate around AI regulation, as articulated by Senator Schiff, signals a significant shift towards demanding accountability for AI development and deployment. The push for companies to retain training data is a direct challenge to the "vacuum it all up" mentality that has characterized early AI growth, aiming to establish a traceable and legally defensible foundation for AI models. This could fundamentally alter the cost and complexity of AI development, especially for models trained on vast, often uncompensated, datasets.

Furthermore, Schiff's concerns about "recursive AI" and the unbridled use of AI in sensitive sectors like defense suggest a growing recognition of AI as a potent national security issue, distinct from mere technological advancement. The comparison to regulating social media indicates a move towards more comprehensive oversight, potentially involving new agencies akin to an "FDA for AI" and a reevaluation of legal doctrines like Chevron deference, which could impact the enforceability of future regulations and expose companies to prolonged legal battles.