Adversarial by Origin How the Classification of External Influence on Machine Meaning Becomes Law Without Becoming Jurisprudence

The paper’s central distinction is:

• harm-based classification: asks what was damaged, who owned it, and whether access or conduct was unauthorized;
• origin-based classification: asks whether the influence came from the platform or from an outside speaker.

The paper argues that origin-based classification can transform ordinary expressive influence into a security event merely because it changes what a machine says.

Five broad claims structure the work:

1. the doctrinal vehicles for formalization already exist; 2. formalization may occur without a complete jurisprudential cycle; 3. rules about machine meaning are especially dangerous when adversarial meaning-making is itself bypassed; 4. the result may create a one-way license for platforms to shape model outputs while public speakers are treated as attackers; 5. intervention must therefore occur outside conventional precedent alone.

The paper preserves a harm-based safe harbor for:

• exfiltration;
• credential theft;
• destructive tool use;
• fraud;
• impersonation;
• unauthorized access.

Its target is not security law as such but the substitution of origin for harm as the classifying variable.