Sam Altman and AI Agents: Linking AI to Digital Identity

OpenAI's New AI Agent and the Future of Online Identity

Sam Altman, CEO of OpenAI, recently unveiled Operator, the company’s inaugural AI agent capable of autonomous action on the internet, marking a significant moment in technological advancement.

However, OpenAI isn't Altman’s sole endeavor focused on leveraging the growing prominence of AI agents.

World Project: Linking AI Agents to Human Personas

Altman’s World project is now developing tools designed to connect specific AI agents with individuals’ online identities.

These tools will enable other users to confirm that an agent is legitimately operating on behalf of a person, as explained by Tiago Sada, the project’s chief product officer.

Formerly known as Worldcoin, World, a web3 project co-founded by Altman and Alex Blania’s Tools for Humanity, operates on the premise that distinguishing between humans and AI agents online will become increasingly difficult.

To address this challenge, World aims to establish digital “proof of human” tools.

Upon scanning an individual’s eye with a specialized silver metal orb, World will issue a unique identifier on the blockchain, verifying their human status.

A History of Pivots and Controversies

It’s important to acknowledge that Tools for Humanity has a track record of strategic shifts and has faced various controversies.

Initially a crypto project, World experienced temporary bans in certain countries and was recently mandated to allow European users to delete their biometric data upon request.

In October, the project removed “coin” from its name, signaling a greater emphasis on human verification rather than cryptocurrency.

A Logical Progression: Verifying AI Agents

While verifying AI agents to perform tasks on behalf of individuals might appear to deviate from World’s core mission of verifying humans with web3 IDs, Sada asserts it’s a crucial and natural progression for the project.

“The ability to delegate your ‘proof of personhood’ to an agent and allow it to act on your behalf is incredibly important,” he stated in an interview with TechCrunch.

“Instead of solely permitting individuals believed to be human onto your website, you’ll also authorize AI agents representing a verified human user. This is where World ID becomes essential.”

Licensing AI Agents and Controlling Online Networks

World’s ID technology could also facilitate the licensing of AI agents to act on a user’s behalf, according to Sada.

A recent blog post from the project highlights that its proof of human tools will not only differentiate humans from bots, but also empower individuals to manage a network of AI agents online.

World is expanding its capabilities to validate “digital actions” associated with a person, even if those actions didn’t originate directly from their account.

The project anticipates its AI-focused tools will be vital in 2025, though Sada indicated the firm will “need to see” if these tools involve linking AI agents to individuals.

However, the startup is clearly considering this possibility.

“In certain applications, it’s irrelevant whether a real person or an agent is initiating the action; the key is knowing a person endorses the interaction,” Sada explained.

Addressing Bot Traffic and Security Concerns

Currently, many websites employ tools from companies like Cloudflare and Snowflake to block AI bots from scraping their content.

Initial testing of OpenAI’s Operator revealed that some websites automatically blocked the new agent.

However, Sada believes businesses may reconsider these restrictions to allow certain AI agents access to their services.

“Ultimately, businesses aim to increase sales and serve a wider audience,” Sada said.

“If a delivery app experiences increased sales due to agents placing orders on behalf of users, the business will likely welcome that.”

Collaborations with Major Platforms

This observation is particularly relevant given OpenAI’s recent partnerships with Uber, Instacart, and DoorDash, enabling its Operator agent to utilize their platforms.

This represents a significant shift for these companies, accustomed to directly serving users through their apps.

Instead, online businesses are now preparing for a future where some user interactions will be mediated by AI agents.

“If an agent presents a verified checkmark and states, ‘I am a bot, but I am authorized to purchase a hot dog for Tiago,’ businesses can grant that agent access,” Sada stated.

Balancing Access and Security

Allowing unrestricted access to all AI agents could expose websites to DDOS attacks, scams, and other malicious activities.

By limiting access to a specific number of AI agents per person, Sada suggests businesses can leverage the benefits of these agents without compromising site security.

A Network of Tools for an AI-Centric Future

Altman’s diverse ventures, when viewed collectively, form a network of tools designed to support a future where AI – and, by extension, OpenAI – plays a central role.

His nuclear fusion startup, Helion Energy, could potentially provide the energy needed to power OpenAI’s data centers, while the longevity science startup he supports, Retro Biosciences, is utilizing OpenAI’s models to explore ways to extend human lifespan.

The Potential for Integration with OpenAI

A crucial question surrounding World is whether its tools will eventually integrate more closely with OpenAI.

Altman regularly participates in World’s events, reportedly engages in weekly discussions with the team, and the project continues to prioritize its AI-related work.

Platforms are increasingly receptive to allowing OpenAI’s agents access to their services, and World may become instrumental in verifying these agents and others.

TechCrunch has an AI-focused newsletter! Sign up here to get it in your inbox every Wednesday.