The new identity paradigm must be hybrid, not just human (www.techradar.com)

As agentic AI and nascent AGI move from research labs into every enterprise app—Gartner predicts AI assistants will be ubiquitous by year-end and task-specific agents will reach 40% of apps by 2026—Anurag Dodeja (Twilio) argues that digital identity must evolve from a human-only gatekeeper into a hybrid, real‑time trust engine that manages both people and autonomous agents. The piece highlights immediate risks (voice/biometric spoofing, synthetic identities) and thorny liability questions—e.g., who pays when an agent overspends or commits expense fraud—making transparency, continuous tracking, and a verifiable “chain of evidence” (think immutable ledgers and protocols like Google’s AP2 or Model Context Protocol) essential for accountability. Technically, Dodeja prescribes abandoning siloed identity silos in favor of a unified platform built on five pillars: continuous verification (beyond passwords/OTPs), bot blocking and agent differentiation, dynamic time‑bound consent and task‑specific authorization, layered behavioral and cryptographic signals for anomaly detection, and persistent account lifecycle memory for seamless personalization. He notes existing tools (privileged access management, multimodal biometrics) can be adapted but need standardized, interoperable frameworks (OpenID, MCP) to enable broad adoption. For AI/ML practitioners and security architects, the implication is clear: design identity systems that monitor intent and context in real time, enforce fine‑grained, revocable agent permissions, and preserve immutable audit trails to balance usability, privacy, and accountability.