Technology should help, not harm.

The pledge

We did not start this one. We signed on to it.

Adding weapons to robots that are widely available and capable of navigating the places where people live and work raises both serious risks of harm and serious ethical objections.

In October 2022, six robotics companies spanning three continents — Boston Dynamics, Agility Robotics, ANYbotics, Clearpath Robotics, Open Robotics and Unitree Robotics — published an open letter to the robotics industry pledging not to weaponize their general-purpose robots or the software that enables their use, and calling on the rest of the industry to do the same.

The Global Genius Trust has made that same pledge. We did not originate it and we claim no credit for it. We have adopted it — and we extend it past mobile robots to every system we develop.

This is not a new position. The Trust and its operating companies have opposed the weaponization of our technologies from the beginning, and our terms and conditions of sale say so explicitly — before a sale, not after it.

What this means in practice

A term of sale, not a statement of intent

An ethical position that lives only on a website is worth very little. Ours is written into the commercial documents:

Across the portfolio

One standard, every project

The Trust does not hold a single product. It holds transport, energy, monetary and foundation work, and the same prohibition applies to all of it — because the risk is not specific to robots. A payment rail, a power system and a food network can each be turned against people just as effectively as a machine can.

Public safety

The same technology, pointed the other way

The reason to refuse weaponization is not that these systems are dangerous. It is that they are powerful, and that power has an obvious better use.

Automated transit and mobile robotics can and should be used as potentially life-saving tools that keep people out of harm’s way — moving people away from hazardous roads, reaching communities that emergency services cannot reach quickly, carrying out inspection and monitoring work that would otherwise put a person in a dangerous place. We support the safe, ethical and effective use of these technologies in public safety. That support and the prohibition on weaponization are the same commitment, stated in two directions.

Why we publish our designs

Open by default is an ethical choice, not only a commercial one

The Trust operates through Open Management, and its engineering follows the same logic: mechanical, civil and electrical designs are published as open source so that any qualified manufacturer can build to them, and so that improvements return to everyone.

This matters ethically, not just commercially. A system whose designs are public can be inspected by the people it affects. It cannot quietly acquire a capability nobody outside the company knows about. It does not depend on one organisation’s continued goodwill, or its continued existence, to keep running safely. Secrecy and safety are often presented as the same thing. In infrastructure they are usually opposites.

Privacy and civil rights

Surveillance is a form of harm

The next generation of these systems will work among us and alongside us more closely and more intelligently than their predecessors. That raises misuse questions that have nothing to do with weapons.

We prohibit the use of our systems in any way that violates privacy or civil rights law, and we will continue to work with policymakers and industry counterparts on the risks that emerging artificial intelligence brings with it — particularly bias, dehumanization, and the erosion of privacy. A transport corridor that watches everyone who uses it is not a neutral piece of infrastructure.

Video and audio monitoring

We will not fix surveillance by pretending we cannot see

Everything on this page applies, without exception, to all video and audio monitoring carried out on or by our systems. A camera on a transport corridor and a microphone in a station are covered by the same prohibitions as any other capability we build: they may not be weaponized, they may not be used to harm or intimidate, and they may not be used in any way that violates privacy or civil rights law.

But refusing to monitor at all would throw away something genuinely valuable. Recorded video and audio is how a collision is reconstructed honestly, how a structural failure is traced to its cause, how an accusation against a passenger or against an operator is settled on evidence rather than on whose word carries more weight. We are not willing to lose that use.

What makes monitoring dangerous is not the recording. It is the asymmetry — one party holds the footage, everyone else is merely in it. So we remove the asymmetry rather than the cameras.

The two conditions exist to protect the people in the recordings, not to gatekeep access to them: an anonymous, unexplained request for footage of a named individual is exactly what a stalker makes. Identification and a stated reason make the request itself part of the record, which is the same principle applied to the person asking.

A monitoring system the public can inspect is an accountability instrument. The identical system, held privately, is a surveillance instrument. The hardware does not decide which one it is — the access rules do.

Building for the future

An invitation to the rest of the industry

We are committed to working with policymakers to promote the safe use of these technologies and to prohibit their misuse, and we call on the broader robotics, transport and automation community to make similar pledges.

We also recognise that our commitment alone is not enough to address every risk. We support this coalition and other ethics initiatives, internally and externally, now and in the future, and we expect this to be the beginning of a long conversation rather than the end of one.

These are powerful systems that help keep people out of harm’s way. We intend to keep them that way, for the safety of generations who had no say in building them.

How we intend to keep this promise

The Sophia

Everything above is a document, and documents do not enforce themselves. Institutions of any kind — charitable, religious, commercial, governmental, even non-profit — tend to become corrupt given sufficient time. A pledge written by the founders is worth exactly as much as the judgement of whoever holds the organisation a generation later.

That is the problem the Sophia exists to solve. Sophia means wisdom, from the Greek sophia. In the Global Genius Trust it is not a metaphor and not a committee: it is a position, held by a qualified person, seated on our projects alongside the researcher, the manager and the financier — with discernment of what is actually at play in a decision, foresight measured in generations rather than quarters, awareness of consequences to the natural world, and the standing to raise an objection that must be answered.

This is how we intend to assure that our technology is never misused. Not by trusting this page, and not by trusting ourselves, but by permanently seating someone whose job is to notice when the answer is drifting — and who has the authority to say so before the drift becomes a decision.

Read The Sophia →