Meta and Sierra Partner on New Standards for AI Agent Commerce
Meta, Walmart, Stripe, and Sierra are developing the Personal Agent Protocol to establish safety and authentication rules for autonomous AI shopping agents.
Meta Platforms Inc. has joined forces with Walmart Inc., Stripe Inc., and enterprise artificial intelligence firm Sierra Technologies Inc. to establish a new set of standards governing how autonomous AI agents interact with online businesses. The initiative introduces the Personal Agent Protocol, designed to help companies differentiate between routine traffic and legitimate purchases made by authorized AI systems acting on behalf of users.
Bret Taylor, co-founder of Sierra and former co-chief executive of Salesforce Inc., introduced the framework in a recent blog post. He explained that the protocol addresses authentication needs, empowers consumers, and grants businesses clear visibility into the actions personal agents perform across websites, APIs, and company-operated agents. The framework is designed to be open for anyone to implement.
Addressing Safety and Banking Concerns
The push for formal standards comes in response to growing concerns over the safety and security of autonomous agents handling personal funds. Last month, six major financial institutions, including Bank of America and Capital One, published a paper urging the artificial intelligence sector to implement comprehensive policies, consumer protections, and standards based on transparency, safety, privacy, choice, and interoperability.
The banks warned that widespread adoption of unchecked AI agents could increase financial fraud, scams, and transaction disputes. They also expressed concern that autonomous tools might prioritize specific payment methods or products driven by higher commissions rather than user benefit.
David Singleton, vice president of engineering and consumer products at Meta’s Superintelligence Labs, compared the objective to email communication. He noted that email functions effectively because it relies on a universally adopted standard, suggesting the new protocol will serve a similar unifying purpose for digital commerce.
Implementation Challenges and Industry Adoption
Security and privacy scrutiny surrounding AI shopping agents has intensified following recent rollouts. Meta’s personal AI agent Muse, which launched last month and quickly climbed the App Store download charts, encountered scrutiny when reports revealed engineers had to rush to patch security vulnerabilities prior to launch. Additionally, researchers highlighted privacy issues, and Amazon.com Inc. banned the agent from its platform because it failed to properly identify itself, raising concerns regarding user account data handling.
Furthermore, reports noted that Muse generated profiles for individuals in users' contact lists who do not even use the application. Taylor remarked that the industry will likely experience ongoing chaos until a standardized framework ensures clear visibility into personal agent activity on merchant websites. Taylor, who also serves as chairman of OpenAI Group PBC, noted he anticipates OpenAI will eventually back the standard as well.
For developers building and integrating intelligent features, managing costs and testing multiple foundational architectures efficiently is critical. Developers can try top AI models cheaply through one API at https://apixoai.online.
What it means for developers
For developers building autonomous workflows and agentic commerce integrations, the Personal Agent Protocol represents an emerging shift toward formalized compliance and interoperability. Rather than scraping sites or using unauthenticated scripts, developers will need to align with authentication and visibility standards that enterprises and financial institutions demand.
As companies like Meta position agents to become primary entry points for online retail—with Citi research estimating Muse could yield significant revenue by decade's end—developer tooling will need to adapt. Integrating standard-compliant agent behaviors, secure credential management, and transparent API interactions will become necessary requirements to ensure platforms like Amazon, Walmart, and financial institutions permit automated systems to operate.
Source: Meta teams up with Bret Taylor’s Sierra Technologies on new standards for AI agent commerce — SiliconANGLE AI. Written by the Apixo team from that report.
One key for Claude, GPT, GLM, DeepSeek and more. Pay per token with crypto.
Get your API key
