IMD AI Safety Clock moves to 23:45 citing agentic AI advances and real‑world integration

IMD points to agentic payments, enterprise always‑on agents, and AI integration into telecom and critical infrastructure as the grounds for tightening its safety clock.

In this brief: 3 sections 2 min read

- IMD states agents now move money—live AI‑executed payments by Santander/Mastercard and Visa across Europe.
- AWS and Google extended infrastructure/protocols to enable agent discovery and payments.
- IMD cites this autonomy shift as central to its move to 23:45.
- Microsoft, OpenAI, and others deployed agentic assistant products for workplace and cloud use.
- Telecom and network vendors are embedding AI agents in network operations and rollout tools.
- IMD flags increased coupling of AI with physical systems (robotics, embodied AI) and military diffusion paths.
- Agentic payments and always‑on agents expand what an attacker or failure can affect (financial, logistics, telecom).
- IMD highlights risks from capability distillation to cheaper local models and edge deployment.
- Recommendation focus: governance of agent permissions, discovery, and human‑present vs human‑absent transaction policies.

Read full analysis on imd.org

Useful?