AI Agents Get Spending Power While Humans Fail Authenticity Tests at 51.2% (syntheticauth.ai)

This week’s biggest identity story is Google’s Agent Payments Protocol (AP2): an open standard, backed by 60+ firms including Mastercard, PayPal and American Express, that lets AI agents spend on users’ behalf via cryptographically signed “mandates.” AP2 records an “intent mandate” (your permission, e.g., “buy tickets under $100”) and a later “cart mandate” (the specific purchase), creating a verifiable chain and non‑repudiable audit trail so purchases can be proven authorized. That capability arrives as humans’ ability to spot AI output collapses—new research shows only 51.2% accuracy—forcing urgent questions about authenticity, authorship and fraud. Practical integrations (Spotify’s persistent agents, Stanford’s BEHAVIOR robotics benchmark) show tangible gains, but corporate missteps—from layoffs justified as “AI‑first” to Meta’s alleged data‑use lawsuits and Harvard Business Review’s “workslop” finding 95% of orgs see no measurable AI ROI—underscore how governance, data practices and human oversight lag technical deployment. At the same time Europe quietly expanded post‑quantum infrastructure with VLQ, a €5M, 24‑qubit system in the Czech Republic that will run in hybrid classical‑quantum mode with the Karolina supercomputer. EuroHPC has procured six varied quantum systems (trapped ions, superconducting, photonics) to hedge approaches; VLQ will be accessible to researchers and industry by 2025. Together, AP2 and emerging quantum capability highlight a paradox: we’re enabling autonomous, auditable digital identities while the foundations of authenticity and cryptographic resilience are being simultaneously stressed and reengineered. Developers, regulators and security teams will need new standards, post‑quantum planning and tighter provenance tools to keep identity trustworthy in an increasingly agentified world.