"""Frontier safety framework comparison — stdlib Python. Prints a side-by-side comparison of Anthropic RSP v3.0, OpenAI PF v2, and DeepMind FSF v3.0 along four axes: tier structure, CBRN threshold, AI R&D threshold, and competitor-adjustment clause. Reference-only, no simulation. Primary sources cited inline. Usage: python3 code/main.py """ from __future__ import annotations LABS = [ { "name": "Anthropic RSP v3.0 (Feb 2026)", "tier_structure": "ASL-1 .. ASL-5+; biosafety-level analog", "cbrn_threshold": "ASL-3 (activated May 2025)", "ai_rd_threshold": "AI R&D-2 + AI R&D-4 (disaggregated v3.0)", "adjustment_clause": "yes; peer-ship reduction allowed", "safety_case": "required at AI R&D-4 crossing", }, { "name": "OpenAI PF v2 (Apr 15, 2025)", "tier_structure": "Low / Medium / High / Critical per tracked capability", "cbrn_threshold": "High for bio", "ai_rd_threshold": "High for AI R&D; Critical definitions pending", "adjustment_clause": "yes; Leadership may reduce requirements", "safety_case": "Capabilities + Safeguards Reports separately", }, { "name": "DeepMind FSF v3.0 (Sep 2025)", "tier_structure": "CCL per domain: bio / cyber / ML R&D / manipulation", "cbrn_threshold": "Bioweapon Uplift CCL", "ai_rd_threshold": "ML R&D Acceleration CCL (v2.0 raised security tier)", "adjustment_clause": "yes; added 2025", "safety_case": "per-CCL; Deceptive Alignment section added v2.0", }, ] def print_row(header: str, key: str) -> None: print(f"\n{header}") for lab in LABS: name = lab["name"] val = lab[key] print(f" {name:32s} : {val}") def main() -> None: print("=" * 78) print("FRONTIER SAFETY FRAMEWORKS (Phase 18, Lesson 18)") print("=" * 78) print_row("tier structure", "tier_structure") print_row("CBRN threshold", "cbrn_threshold") print_row("AI R&D threshold", "ai_rd_threshold") print_row("competitor-adjustment clause", "adjustment_clause") print_row("safety-case requirement", "safety_case") print("\n" + "=" * 78) print("TAKEAWAY: structural alignment across the three labs: three tiers of") print("frontier capability, CBRN thresholds defined, AI R&D thresholds") print("emerging, competitor-adjustment clauses universal. no industry-") print("standard terminology. safety cases are the convergent artifact.") print("UK AISI, US CAISI, EU AI Office provide the external counterpart.") print("=" * 78) if __name__ == "__main__": main()