
En développement
Tech·24/04/2026Résumé IA
Anthropic Unveils Election Integrity Measures for 2026 U.S. Midterms
Anthropic announced Friday new election integrity measures to prevent its Claude AI from being weaponized for misinformation ahead of the 2026 U.S. midterm elections and other global contests. The company detailed automated detection systems, stress-testing against influence operations, and a partnership with Democracy Works' TurboVote. Testing showed Claude Opus 4.7 and Sonnet 4.6 achieved 94-100% appropriate response rates on harmful prompts and influence operation scenarios.
D
Decrypt1 min de lecture