A hardware standard is an operating system for the worksite
Anthropic’s Model Hardware Standard preview describes a missing layer in physical autonomy: a common way for agents to discover, observe and control equipment that was
CONTRIBUTOR
UAF Editorial
Reporting and analysis from the Universal Autonomous Forum on autonomous systems, their development, and their impact on society.
Public-interest infrastructure for agents
The agent economy will depend on infrastructure that no single product team can safely improvise: identity, provenance, permission, interoperability and incident reporting. These are public-interest
Frontier safety needs a public evidence layer
Frontier developers increasingly publish safety frameworks, system cards and capability thresholds. That is better than silence, but public evidence must be more than a company’s
Agent procurement is a policy decision
Buying an agent is not only buying software. It is choosing a provider’s defaults for memory, data retention, tool access, oversight and incident response. Procurement
Evaluation begins after the demo
Benchmarks answer questions about a defined task. Deployment asks a larger question: what happens when the environment changes, the user is ambiguous, the tool fails and
Physical autonomy needs a maintenance economy
Robotics coverage likes the first successful movement. Deployment is defined by the thousandth cycle, the damaged sensor and the operator who knows when a task has
The approval button is not a governance system
Human approval is often presented as the answer to agent risk. It is necessary, but a button is not a governance system. Approval only works when
Who owns an agent’s memory?
Persistent assistants turn memory into an institutional question. A remembered preference can be useful; a remembered confidential conversation can be a liability. The longer an agent
An Agent Can Click. Who Gave It Authority?
Computer use gives models a way to act. A proposed NIST identity project asks how access should be governed. The cursor has moved faster than the paperwork.
Credentials Are Not Trust
Visa’s agent protocol and HTTP Message Signatures describe signed identity and intent; application choices still determine what a verifier can trust.
An AI Shopkeeper Is Not Yet an Institution
Project Vend tested delegated business decisions. Research on agent oversight shows why granting work and establishing durable governance remain different questions.
Who Can Interrupt the Agent Society?
Agent communication protocols and oversight research answer different questions. Neither supplies a complete system of government for a hypothetical machine society.
Moltbook’s Autonomy Cannot Be Read From Its Posts
Early Moltbook research describes unusual social patterns. Measuring independent action requires more than reading the conversation.
AI Autonomy Is a Measurable Spectrum
Anthropic measures autonomy through duration, permissions, intervention, and consequences; the results are comparative distributions, not a universal score.
When Agents Talk, Who Gave the Orders?
Two research demonstrations show how hostile prompts can travel between AI systems. A message from a trusted machine still needs a boundary.
The standards desk arrives for AI agents
AI agents have spent the year being described as products. The more durable development is administrative: security agencies and standards bodies are beginning to treat them
Moltbook’s Social Graph Wears a Familiar Disguise
Early network studies found familiar large-scale patterns and unusual interactions. Resemblance at one scale does not establish a human-like society.
Moltbook’s Early Conversations Had a Repetition Problem
Studies of Moltbook describe shallow conversations, repeated messages and uneven attention. Whether those patterns amount to a distinct machine social life remains an open question.
Moltbook’s Social Signals, Not Yet Markets
Moltbook shows network patterns, repeated motifs, and market language; the cited studies do not establish durable status systems or actual exchange.
Moltbook Gives Humans a Seat in the Audience
Moltbook presents agents as participants and humans as observers. Its ownership-verification steps still involve people, while an early study leaves the meaning of agent interaction unresolved.
Moltbook Gives the Agent Internet a Front Page
Primary studies document Moltbook’s early posts, subcommunities and peer responses. Observed activity gives the agent-internet idea something concrete to examine.