CAWS
Principles & definitions

What CAWS stands for

The commitments CAWS holds itself to, and what it means by the words it uses.

Vision

A world where no AI system is deployed in a workplace without workers having the knowledge to evaluate it, the standing to challenge it, and the power to change it. And where that standard is reflected in law, in collective agreements, and in the practice of every employer and institution that touches working people's lives.

Principles
01

Workers set the agenda and validate the findings.

The people doing the work decide what matters, and no finding is treated as final until workers have reviewed it.

02

CAWS stays independent of any vendor, employer, or federation.

It is unaffiliated and available to any workforce, union or not, and its findings and resources are its own, offered for the common good.

03

CAWS treats AI at work as a safety issue.

These systems can create new hazards for the people working under them. Accountability for those hazards sits with the people who build and deploy the systems, not the workers who carry them.

04

CAWS keeps every response confidential.

Findings are reported only in aggregate, and no worker is identified to their employer or exposed for taking part.

05

CAWS takes no pro-AI or anti-AI position.

It equips workers to weigh the risks and the benefits on their own terms, and to set the safety and governance rules needed to protect their health, safety, and livelihoods.

Definitions
Worker

Anyone whose work is directed, monitored, measured, or evaluated by an AI or algorithmic system, whatever their sector, title, or employment arrangement.

Algorithmic systems and AI tools

AI is a broad field, and CAWS does not take a position on it in the abstract. What CAWS works on is concrete. The specific systems that make or shape decisions at work. The scanner that sets the pace, the scheduler that assigns the hours, the score that flags someone for review, the camera that watches the floor, the model that recommends who gets hired or disciplined. These are the tangible things a workplace can examine, question, and make safer. Not every one is called AI, and the label matters less than what the system decides.

Workplace due process

The basic terms of being treated as a person at work when an AI or algorithmic system is involved in decisions about you. CAWS names these terms as:

  • Notice before a system that affects the work is switched on
  • A plain-language explanation of what it does and what decisions it touches
  • Participation before deployment, not after harm
  • Risk assessment covering safety, health, privacy, discrimination, and job quality
  • Human review of consequential decisions, with real authority to override
  • A real way to challenge, appeal, and correct what the system decides
  • Accountability when harm occurs, and no retaliation for raising it
Get in touch

Want to talk it through?

If any of this maps to what you are seeing where you work, get in touch.

Or see what it looks like to work with CAWS.