The news
Meta Platforms (META) and Sierra Technologies, the AI customer-service company co-founded by Bret Taylor, on October 6 introduced the Personal Agent Protocol, a proposed standard for how AI agents that act for consumers interact with businesses online, SiliconANGLE reported.
According to SiliconANGLE, the protocol covers how agents authenticate, how consumers stay in control, and how businesses get visibility into what agents do across websites, APIs and company systems. Walmart (WMT) and Stripe were named as involved. Taylor, who is also OpenAI's chairman, said the standard is open for anyone to implement.
David Singleton, a vice president at Meta's Superintelligence Labs, compared the effort to email, which works because everyone follows one standard. The outlet described the move as a response to a call from six major banks for agent standards covering transparency, safety, privacy, choice and interoperability.
The launch lands while many large sites are keeping agents out. TechCrunch reported on October 6 that Amazon (AMZN) has explicitly blocked Meta's Muse agent from its retail site, and that Walmart's human-verification checks have unintentionally blocked agents. United Airlines (UAL) restricts automated access, Yelp (YELP) requires paid data-licensing deals, and eBay (EBAY) restricts unauthorized agents, with some users reporting account suspensions, according to TechCrunch.
TechCrunch also reported that Cloudflare changed crawler defaults on September 15 in a way that blocks AI agents on pages carrying ads. Meta's position, as quoted by TechCrunch, is that turning away a personal agent means turning away the customer behind it.
The numbers
- AI-assisted purchases closed on Amazon
- 59.7% (PYMNTS Intelligence)
- U.S. consumers starting shopping research with AI
- 61 million (23%)
- Completed on brand websites or apps
- 6.3%
- Banks that called for agent standards
- 6
Why CEOs should care
For retail and e-commerce leaders, agent access is no longer a bot-management setting buried with the security team. Blocking agents protects pricing data and checkout from abuse, but it may also turn away a paying customer who delegated the purchase. Ask your team which agents currently reach your site, which are blocked by default rules you never chose, and whether your CDN's September changes altered that.
CFOs should look at where the sale actually lands. PYMNTS Intelligence data, from an August survey of 2,191 consumers, found 59.7% of AI-assisted purchases closed on Amazon and only 6.3% on brand websites or apps. If agents become the front door, brands that are hard for agents to transact with risk ceding more volume to marketplaces. Walmart executive Daniel Danker described the risk of handing customers to AI platforms as intermediation, as quoted by PYMNTS.
CISOs and boards need a policy, not just a firewall rule: how agents authenticate, what they may see (prices, inventory, account data), what they may buy, and who is liable when an agent makes a mistake. The Personal Agent Protocol is a proposal, not an adopted industry rule, so evaluate it alongside existing terms of service and fraud controls rather than as a ready answer.
The bigger picture
Agents are arriving faster than the rules. Meta's Muse and other agents already try to browse and buy, while sites set access policy one by one. A shared protocol backed by Meta, Sierra, Walmart and Stripe could give businesses a common way to identify and control agents, but rival standards and holdouts such as Amazon could leave the market fragmented for some time.
What’s next
Watch whether other retailers, payment networks and agent makers sign on to the Personal Agent Protocol, whether Amazon changes its stance on Muse, and whether airlines and marketplaces loosen terms that bar automated access without written permission.
What “Fact-checked” means
Fact-checking means testing a story’s facts against the evidence before it is published. This story went through at least two separate checks before this version was published.
- What we checked
- Its names, figures, dates, job titles, quotes and who said what were checked against the story’s sources, including its main source where it could be opened. The headline was checked for accuracy and overstatement.
- How
- A first check reviewed the whole story. If it passed, a second, skeptical check went back to the sources to look for mistakes in the most important facts. If a check flagged the story, it was edited to fix the problems found, and a separate re-check then reviewed the whole story again.
- Who
- The checks are made by our newsroom, as steps kept separate from the writing, under rules set by our editor, Hussein Mukhtar. A story the checks still flag is held for the editor, who decides whether it is fixed, published or dropped.
- If something is wrong
- “Fact-checked” does not mean error-free. If a material error is found after publication, we correct the story and add a note saying what changed. Report an error
