AI agents acting on behalf of users are becoming more common, influencing how businesses must adapt their digital platforms. These agents can perform tasks like shopping, booking, or managing accounts, raising unique challenges around identity, security, and trust.

Current Business Approaches to AI Agent Access

Businesses are adopting different policies for letting AI agents interact with their systems. JPMorgan Chase has chosen a restrictive approach, blocking access to prevent unauthorized agent activity, as CEO Jamie Dimon indicated during a recent earnings call. Other companies like Charles Schwab and Delta Airlines are similarly cautious, restricting AI agents to protect user data and system integrity.

Conversely, Bank of America has positioned the responsibility on customers, advising them against sharing login credentials with AI agents but warning that any activity through shared access is the customer’s accountability. This approach highlights the risks users face when granting AI agents direct access without safeguards.

Walmart is pioneering a more nuanced solution with its introduction of Agent Profiles, a dedicated framework within customer accounts that allows AI agents to act on the customer’s behalf under regulated conditions. By giving each AI agent a separate profile, Walmart ensures transparency and control over agent activities, differentiating between the user’s direct actions and those performed by AI.

The Significance of Separate Agent Profiles

With separate profiles, businesses can track and manage AI agent behavior distinctly from the human user. Walmart’s system enables setting spending limits and requires approval for purchases, mitigating risks of unwanted transactions. Agents also receive the same pricing information as customers, maintaining fairness and consistency.

This model recognizes the unique nature of AI agents who may browse many more pages or make decisions the human user might not consider, all while acting with the user’s authority. Differentiating agents from users supports better analytics, fraud prevention, and customer experience.

Technical and Security Considerations

AI agents generally operate by signing in with a customer’s credentials, which raises concerns about identity verification and authorization. Emerging specifications, such as the Personal Agent Protocol currently in development by over 40 companies including Meta, aim to standardize how AI agents identify themselves to platforms.

This protocol recommends agents use methods like Web Bot Authentication to prove which organization powers them, allowing businesses to validate and monitor agent activity. Some AI agents, like Meta’s Muse, have already started signing requests with such identities without public announcement.

Security vendors and website operators must evaluate their infrastructure’s ability to distinguish AI agents from humans. Automated bot defenses might inadvertently block beneficial AI traffic or let malicious agents slip through. Preparing systems for agent-aware access controls is critical as agent usage scales.

Industry Impact and User Perspectives

Companies like DoorDash have reported only a few hundred AI agent orders so far within hundreds of millions of total transactions, suggesting that large-scale AI agent adoption is still in its early phases. This window allows businesses to design policies before AI agents become mainstream online users.

“Businesses that proactively implement agent-aware systems will establish trust and competitive advantage as AI agents proliferate,” stated a cybersecurity analyst at a leading consultancy.

The ability to control agent actions through profiles or spending limits may become essential for protecting customers and platforms alike. Customers also benefit from clear visibility on what their AI agents are authorized to do, reinforcing trust in these new digital helpers.

Balancing User Convenience and Security

The challenge lies in enabling AI agents to efficiently perform tasks without compromising security or user control. Systems like 1Password for Agents demonstrate how password managers can securely relay login information without exposing credentials to AI agents, further enhancing protection.

Adopting agent profiles or authentication protocols can mitigate risks associated with account sharing or unauthorized access, providing a more scalable framework than outright blocking agents.

For businesses unsure of the best approach, it is advisable to conduct thorough assessments of their bot management policies and prepare to create agent-specific access controls tailored to their operational and customer needs.

Preparing for the AI Agent Era

Businesses must take proactive steps to adapt to AI agents. They should assess how current security and bot filtering tools interact with AI-driven traffic, which can be indistinguishable from human activity or exploit existing vulnerabilities.

Deciding on a clear policy regarding AI agents—whether blocking them, delegating responsibility to customers, or implementing agent profiles—allows businesses to maintain control and protect their customers. Monitoring and updating systems to recognize signed AI agent requests will soon become a critical operational capability.

For companies seeking to leverage AI agents for improved automation and customer engagement, integrating AI-friendly features is important. Tools that enable secure authorization and transparent tracking of AI agent activity can unlock new levels of service innovation while maintaining compliance and risk management.

Those interested in how AI agents influence paid media and ad performance can explore the diagnostic approaches to Meta ad campaigns demonstrating AI’s broader marketing impact. Additionally, businesses can learn from strategies to generate demand with measurable results using AI-based automation tools.

Conclusion

The evolving presence of AI agents on digital platforms challenges traditional notions of user interaction and security. Businesses like Walmart are leading the way with innovative solutions such as separate agent profiles, balancing convenience and risk mitigation.

Others favor stricter controls or place responsibility on the user, but overall, readiness and adaptability will dictate future success. Establishing standards and protocols for agent identification, as well as improving security layers, can enable safe and efficient AI agent engagement.

For enterprises looking to stay ahead, adopting agent-aware systems and exploring automated solutions like those provided by Adsroid’s AI-powered marketing platform can facilitate integration of AI agents while safeguarding customer trust.

As AI agents increasingly shape e-commerce and digital services, businesses that understand these dynamics will be positioned to innovate and lead in the new era of agentic commerce.