Proposes detecting large AI training runs from datacentre power draw via Viterbi tracking of iteration cadence, showing it is robust to most operator hiding strategies and useful for compute verification under AI governance agreements.
Shows an external power-draw side channel can identify AI training versus inference on an H200 GPU for compute governance, and a hardened detector still catches most evasion attempts.
Proposes concrete domestic US policies to pace frontier AI development, including compute allocation minimums for safety research, caps on AI R&D use, and safety-case-based existential risk thresholds enforced by third-party assessors.
Argues that an international AI slowdown can be verified with human inspectors given whole-lab access to frontier labs, and that the technical verification barrier is not a reason to delay it.
Critiques the AI Futures Project's Plan A pause-and-compute strategy, arguing that accumulated compute and software overhang (dry tinder) could make a deal breakdown far more dangerous, and that mutually assured compute destruction is fragile.
Examines whether Trusted Execution Environments can give verifiable, privacy-preserving enforcement and monitoring of AI deployments for governance, finding hardware auditability to be the main obstacle.
Evaluates whether GPU telemetry can detect hidden ML training under adversarial evasion, testing a classifier against 20 evasion strategies to support compute governance verification.
Analyzes whether inference-verification measures like bandwidth limits, memory wipes and proof of work could let governments quickly slow AI training, concluding current prototypes are likely ineffective but some measures might buy time.
A shallow prioritization post arguing that neglected infosec work, such as securing model weights, authorization for misuse, compute verification, and controlling untrusted AI labor and rogue deployments, is high-impact for reducing existential risk from AI.
Proposes a global compute governance framework, a 'Compute Pause Button', using tamper-proof FLOP caps, chip traceability and export controls to prevent training of dangerously powerful AI systems.
Report analyzes how flexible hardware-enabled guarantees could support international AI governance on proliferation, loss of control, military AI, and strategic stability, with game-theoretic stability analysis.
Presents the Superintelligence Strategy framework proposing Mutual Assured AI Malfunction deterrence, nonproliferation via compute security, and national competitiveness measures to manage catastrophic superintelligence risks.
Argues the inference-time scaling paradigm (o1/o3) reshapes AI safety: reduced deployment overhang, more useful chain-of-thought oversight if non-language CoT is banned, easier-to-steal but costlier models, and export control implications.
Argues compute providers should act as securers, record keepers, verifiers and enforcers in AI regulation, and analyzes technical feasibility of targeted, privacy-conscious compute oversight under EO 14110 and the EU AI Act.