AI News

Anthropic's Lobbying Spend Surges from $720,000 to $8.11 Million in Two Years: Inside the Washington Power Game of an AI Safety Standard-Bearer

Federal lobbying disclosures show that Anthropic's federal lobbying spending rose from $720,000 in 2024 to more than $3.1 million in 2025 and $8.11 million in 2026, with contracts involving people linked to Secretary of State Marco Rubio and White House Chief of Staff Susie Wiles. The spending spans export controls, defense AI procurement, and acceptable-use rules, highlighting the tension between Anthropic's AI safety narrative and its commercial and policy interests.

Anthropic AI政策 芯片出口管制
75

WDCD Five-Scenario Side-by-Side Review: Business Rules Floor Score Only 1.3, 15-Model Uneven-Performance Gap Reaches 2.7 Points

In the WDCD v3.1 pilot, Business Rules and Security Compliance both have a floor score of 1.3/4, the lowest among constraint types, while model performance varies sharply across scenarios, with gaps of up to 2.7 points. The results point to clear scenario-specific strengths and weaknesses that affect enterprise model selection.

WDCD Compliance Test 场景横评
80