Meta Wants to Write the Industry Standards for Safe Agentic Commerce, Report Says

Meta is reportedly teaming up with Walmart, Stripe, and enterprise AI startup Sierra to publish a set of standards for how AI agents should interact with businesses online. Called a “personal agent protocol,” the standards will reportedly help companies distinguish authorized personal agent purchases from others. “We’re defining rails that we hope personal agents and business agents can run over for the future,” Meta Superintelligence Labs vice president of engineering and consumer products, David Singleton, told CNBC. “If you think about email, it works great because it is a standard that everyone can use to talk to each other, and we think this is going to be quite similar.” AI developers, as well as some online retailers and payment processors, have been betting big on agentic commerce. But there are still widespread concerns over potential safety and privacy problems that arise from giving AI agents access to your entire fortune and then letting them loose on the internet. Last month, a group of six major global banks, including Capital One and Bank of America, released a paper, saying that agentic commerce needed to establish industry-wide standards, policies and consumer protections around five key principles —transparency, safety, privacy and data, choice, and interoperability— before aiming for more widespread acceptance. A rise in agentic commerce could raise the rates of scams, fraud and disputes, the report claimed, and the agents could prioritize certain products or payment methods simply over incentives like higher commissions rather than going for the customer’s needs. Privacy concerns have also specifically been an issue for Meta’s new personal AI agent Muse, which the tech giant launched on Sept. 8 equipped with agentic commerce capabilities. A 404 Media report from Monday claims that Meta engineers had to work overtime to patch multiple severe security vulnerabilities before launch day. Despite that, researchers have found security and privacy issues since the launch. Last month, Amazon locked Muse agents out of its online shopping platform, saying that the agents’ failure to identify themselves on Amazon.com raised concerns over the handling of customer credentials and account data. Then, this week, the AI agent made headlines with yet another report over alleged concerning privacy practices. The AI agent creates and updates “a page for every person in the user’s life” hourly, according to a Wired report. Sierra co-founder Bret Taylor, who leads the formation of the protocol, told CNBC on Tuesday that it will be “chaos until such a standard exists” to provide visibility into what personal agents do on these businesses’ websites. Taylor is also chairman of OpenAI, which unveiled its own personal AI agent offering, and he told CNBC that he expects the AI giant to eventually participate in the standards as well. Meta is also reportedly hoping for the standard to help make the use of AI commerce agents smoother for customers, which would help with wider adoption. Also on Tuesday, Citigroup estimated that Muse could drive user adoption strong enough to generate upwards of $27 billion in revenue before the end of the decade, claiming that it is becoming “the front door to the internet” for online shopping, per a Reuters report.