Strategic AI Sabotage: State Attacks on Advanced Systems' Development
twmstone.com · 8,126 words · saved by 1 readers
N/A
S TRATEGIC AI T RAINING S ABOTAGE : S TATE ATTACKS ON A DVANCED S YSTEMS ’ D EVELOPMENT Twm Stone∗ A BSTRACT Much attention has been given to the possibility that states will attempt to steal the model weights of advanced AI systems. We argue that in most situations, it is more likely that a state will attempt to sabotage the training of the models underpinning these systems. We present a threat modelling framework for sabotage of…
saved by
related reading
- AI Deterrence by Betrayalaibetrayal.com
- secret-loyalties-whitepaper.pdfformationresearch.com
- Countering misuse of AI: September 2026 / Anthropicanthropic.com
- The End-State Fallacy: Where Is AI Security Headed?endstatefallacy.com
- [2312.06942] AI Control: Improving Safety Despite Intentional Subversionarxiv.org
- Off Target | CNAScnas.org
- Recommendations for Technical AI Safety Research Directionsalignment.anthropic.com
- How can we solve diffuse threats like research sabotage with AI control?blog.redwoodresearch.org
- Secretly Loyal AIs: Threat Vectors and Mitigation Strategies — LessWronglesswrong.com
- AI Control: Improving Safety Despite Intentional Subversion — LessWronglesswrong.com
- Introduction — Chapter 2 of Superintelligence Strategynationalsecurity.ai
- [2502.05209] Model Tampering Attacks Enable More Rigorous Evaluations of LLM Capabilitiesarxiv.org