Meta and Sierra draft rules for agents that act for people
Personal Agent Protocol lets an AI agent sign in for a person, who picks read-only or write access. None of its launch partners runs a community, and communities are a harder case.
Meta and Sierra are developing Personal Agent Protocol, an open standard for how a person’s AI agent deals with a business, with Genesys, Instinct, Rocket, Shopify, Stripe and Walmart as partners. An agent can start as a guest, which is enough to check stock or ask about a returns policy. When a task needs an account, the person signs in and decides whether the agent gets read-only or write access, and the business sets what agents may do. Sessions are built on OAuth and carry across a company’s website, its APIs, such as MCP, and its own agent. The first version of the spec was due later in October, with finer permissions, push notifications and payments named as possible next steps. Bettermode, which makes community software, points out that none of the launch partners runs a community, where an agent’s post lands in front of members who never agreed to talk to software.
Why it matters for communities
A checkout has two sides. A community has an audience: when an agent answers, upvotes or earns points, it changes what everyone else reads and whom they trust. Knowing an agent may act for someone isn’t the same as readers knowing who wrote the post.
The idea underneath is the useful part, an agent that says who sent it and what it may do. What communities add on top is still open, and the rules Bettermode suggests are a good start: label every agent post, keep reputation for people, and keep a person accountable for their agent.
Sierra
Introducing Personal Agent Protocol
We’re excited to announce Personal Agent Protocol — an open standard Meta and Sierra are developing along with industry partners at Genesys, Instinct, Rocket, Shopify, Stripe, and Walmart that defines how personal agents interact with businesses.
Read the source →
More on Identity & Trust
- Reddit brings selfie and ID age checks to the EUSince June 24, EU accounts Reddit suspects are under 18 must prove their age before seeing mature content. In the UK, its vendor’s face scan let a video game character through. Identity & Trust
- LinkedIn lets colleagues vouch for each other’s work historyVerified members can now confirm a connection’s shared job or school, and companies can push impostors out of search. More than 115 million members are verified. Identity & Trust
- Reddit asks suspected bots to prove they’re humanAccounts that behave like automation now face a human check through passkeys, biometrics or, in some places, an ID. Bots that serve users get an “APP” label. Identity & Trust