Policy paper: four strategies to boost industry-wide AI safety cooperation

AI Impact Summary

The policy work identifies four actionable approaches for elevating safety norms in AI development, addressing incentive misalignment created by competitive pressure. If adopted, these strategies can reduce under-investment in safety by increasing visibility of risks and benefits, enabling cross-company collaboration without sacrificing competitive advantage. Practical steps for engineering leadership include forming joint risk assessment initiatives, shared threat models, enhanced transparency practices, and a framework to reward adherence to common safety standards.