Institute for Strategic Dialogue
institute-for-strategic-dialogue-29e1e6dd·3 events·first seen 15d agoAliases: Institute for Strategic Dialogue
Co-occurring entities
More like this (12)
Recent events (3)
Anthropic Details Claude Safeguards Team Structure and Multi-Layer Safety Approach
Anthropic has published a detailed overview of its internal Safeguards team, describing a multi-layer approach to preventing Claude misuse that spans policy development, model training influence, pre-deployment evaluation, and real-time enforcement. The team uses a Unified Harm Framework covering five dimensions (physical, psychological, economic, societal, autonomy) and conducts Policy Vulnerability Testing with external domain experts in areas like terrorism, child safety, and mental health. Pre-deployment evaluations include safety assessments, CBRNE-focused AI capability uplift testing with government partners, and bias evaluations. The post describes specific partnerships with organizations like the Institute for Strategic Dialogue and ThroughLine to inform election integrity and mental health response policies.
Anthropic publishes elections-risk testing methodology and releases automated evaluation tools
Anthropic describes its two-stage process for identifying and mitigating elections-related risks in Claude: qualitative 'Policy Vulnerability Testing' (PVT) conducted with external subject matter experts, followed by large-scale automated evaluations. The post details how findings from PVT inform mitigation strategies such as policy updates, model fine-tuning, and response behavior changes, with a case study on election administration accuracy. Anthropic is also releasing some of its automated evaluation tools publicly to help other organizations improve election integrity efforts.
Anthropic details red teaming methods and calls for standardized AI testing practices
Anthropic published a detailed overview of red teaming approaches used to test Claude and other AI systems, covering domain-specific expert testing, automated red teaming, multilingual/multicultural testing, and multimodal red teaming. The post documents empirical findings about when each method is appropriate, highlights partnerships with organizations like Thorn, Institute for Strategic Dialogue, and Singapore's IMDA, and closes with policy recommendations for building a standardized AI testing ecosystem. The piece is notable for its operational specificity and its explicit call for industry-wide standards to enable cross-system safety comparisons.