[Opinion] Welcome to the feudal states of AI

1) Centralisation is freedom. "Congress should preempt state AI laws" & "States should not be permitted to regulate AI development". A single national standard optimised for "AI dominance" overrides local protections, ethics, and dissent i.e. faster deployment of harmful systems with reduced democratic control.

2) Control is accountability. "Congress should not create any new federal rulemaking body" & "industry-led standards". The entities building the most powerful systems also define the limits of acceptable behaviour i.e. regulatory capture by design.

3) Participation is duty. "make federal datasets accessible" & "training of AI models on copyrighted material does not violate copyright laws". Everything becomes training data whether public, private, creative or personal. Extraction of value without compensation, creating a permanent asymmetry between citizens and AI firms.

4) Exclusion is resilience. "streamline federal permitting for AI infrastructure" & "behind-the-meter power generation". Parallel infrastructure systems optimised for AI, not citizens. In effect, resource concentration (land, energy and water) with local communities carrying externalities e.g. pollution.

5) Secrecy is security. "national security enterprise ... to understand frontier AI" & "consultation with frontier AI model developers". AI becomes part of a permanent national security apparatus. Another justification for further secrecy in the AI space.

6) Truth is alignment. "prevent the United States government from coercing technology providers". Platforms (and AI systems) become the primary arbiters of information and truth with minimal external constraint. There will be no democratic oversight of the algorithmic shaping of reality!

7) Instability is opportunity. "task-level workforce realignment driven by AI" & "incorporate AI training... workforce programs". Continuous displacement of workers becomes normal, with perpetual reskilling to chase a moving target. In short, a future with structural dependence on AI systems with workforce protection replaced with adaptation.

8) Surveillance is safety. "commercially reasonable ... age-assurance requirements" & "avoid setting ambiguous standards about permissible content, or open-ended liability". Broader identity infrastructure under the banner of safety. Think normalised identity verification, reduced anonymity, expanded surveillance, limited liability for harms caused and reduced judicial discretion.

In short, a constant trade from local control to centralisation, from regulation to acceleration, from protection to adaptation, from ownership to access and from public oversight to industry governance. Future citizens are reduced to data sources that are constantly monitored and require continuous retraining.

Well done Trump! George Orwell couldn't have done better. This is a masterclass. Freedom is slavery in the land of the free.

Have some slop to replace the front cover.

AI-generated illustration of the feudal states of AI

Originally published on LinkedIn.