Search
The Agentic ID Challenge
How can we assign reliable identities to non-human AI agents?
The Agentic ID Challenge

Agentic AI is the next big evolution of the AI wave. It promises to change the way people and organisations get things done. Deloitte reckons half of employers will try out agentic AI by 2027. Bank of America Global Research says spending on the tech could reach $155 billion by 2030.

The hype is huge. But one critical facet of agentic AI is often overlooked: identity.

Here’s the problem. agentic AI challenges the fundamentals of ID and authentication in digital spaces. Historically, every ID system had a human dimension. But with agentic AI, there’s no person present to type in a user name, scan a passport or blink into the camera.

As agents move into the consumer space, their actions must be traceable back to a verified, consenting human. Without good guard rails, ‘bad actor’ agents will be able to perform account takeovers, sabotage, ransom scams and so on. For the MEF community, this is a highly important issue since so many MEF members are engaged in matters of identity and authentication.

Hence this positioning paper. It explores:

• What is agentic AI?
• Market size and projections
• Major players, product launches and protocols
• Technology and standards
• The challenge of agentic AI identity
• ‘Rules’ for agentic AI ID

At present, we are still in the early stages of defining how agentic ID might work. As the industry investigates possible solutions, this paper provides an important introduction to this critical space.

Key Takeaways
Agentic AI is the next big tech wave.
Bank of America Global Research says spending on the tech could reach $155 billion by 2030.
Agentic AI the next phase of generative AI.
Agents are like digital executive assistants. They can set goals, and act on them without continuous human prompts.
Agentic AI challenges all the fundamentals of ID and authentication in digital spaces.
When an agent carries out a task, there’s no human present to create a password, type in a passcode or smile at the camera.
Any ID system for agents must answer question such as:
Who built this agent? Who owns it right now? What’s its historical behaviour? Has it ever been compromised? Should I approve this request?
A small number of specialists are already trying to build Know Your Agent (KYA) solutions.
However, we are still in the early stages of defining how agentic ID might work.
Telcos can play a key role in this space.
Silent authentication provided by network APIs can reliably identify a device owner without the need for human action.
Sign in to download this report

Apply to speak at:

The Agentic ID Challenge

Apply to speak at: