35 Frameworks. Why That Number Is Not Arbitrary.

When people first encounter Alt-TAB, the free pre-deployment AI ethics assessment tool built on the Parallax methodology, one of the first questions is about the number. Why 35 frameworks? Why not 10? Why not 100?

The answer is not that 35 is a round number or a target that was set in advance. It is that 35 is the number of distinct international governance frameworks that, taken together, provide adequate coverage of the analytical territory that a pre-deployment assessment needs to cover.

Understanding why requires understanding what a governance framework actually is, and what it is for.

A governance framework is not a rulebook. It is a structured articulation of values, principles, and analytical approaches that a particular community, a standards body, an international organisation, a professional association, a government, has developed to guide decisions about technology.

Each framework represents a distinct tradition of thinking about what technology does to people and institutions. The IEEE's approach to AI ethics reflects the values and analytical methods of an engineering standards community. The UN's approach reflects the values and analytical methods of an international human rights system. The OECD's approach reflects the values and analytical methods of an economic policy community.

These are not the same tradition. They do not always agree. And that disagreement is not a problem to be resolved. It is information.

A single governance framework, however well-designed, reflects the values and blind spots of the community that produced it. It will be thorough in the areas that community cares about and has expertise in. It will be less thorough, or silent, in the areas that community has not prioritised or does not have the expertise to address.

This is not a criticism of any particular framework. It is a structural feature of how frameworks are produced. They are produced by communities, and communities have perspectives.

The consequence for assessment is that using a single framework will produce findings that are thorough within that framework's scope and silent outside it. If the risks that matter for a particular technology decision fall outside the framework's scope, they will not be found.

The only way to address this is to use multiple frameworks, specifically frameworks that represent different analytical traditions, different value systems, and different areas of expertise. The goal is not consensus. The goal is coverage.

The 35 frameworks in Alt-TAB were selected to provide coverage across several distinct dimensions.

Analytical tradition. The frameworks include contributions from engineering standards bodies, international human rights organisations, economic policy institutions, professional associations, and national regulatory frameworks. Each tradition brings different analytical tools and different questions.

Geographic scope. The frameworks include contributions from international bodies, regional bodies (including the European Union and the African Union), and national frameworks from jurisdictions with distinct regulatory approaches. This matters because technology does not respect borders, and the harms it can cause are not uniformly distributed.

Domain expertise. The frameworks include contributions from communities with deep expertise in specific domains: child safety, gender-based violence, cybersecurity, labour rights, environmental impact, and others. Domain expertise matters because the risks in each domain are specific, and generic frameworks will miss them.

Temporal coverage. The frameworks include both established standards that have been tested over time and more recent frameworks that address emerging risks. The field is moving quickly, and an assessment that relies only on established frameworks will miss risks that have only recently been named.

Three domains receive mandatory screening in every Alt-TAB assessment, regardless of the technology type: technology-facilitated gender-based violence, child safety, and cybersecurity.

These are mandatory not because every technology is a risk in these domains, but because the consequences of a missed finding in these domains are severe enough that the question should always be asked. The cost of asking and finding nothing is low. The cost of not asking and missing something is potentially catastrophic.

This is a methodological choice, not a political one. It reflects a considered judgment about where the asymmetry between the cost of asking and the cost of missing is most pronounced.

The value of running a technology decision through 35 frameworks simultaneously is not the sum of 35 individual findings. It is the pattern that emerges from the relationships between those findings.

When multiple frameworks from different analytical traditions converge on the same concern, that convergence is significant. It suggests that the concern is not an artefact of one tradition's particular preoccupations. It is something that multiple distinct ways of thinking about technology have independently identified as important.

When frameworks diverge, when one tradition identifies a risk that others do not, that divergence is also significant. It may indicate that the risk is real but not yet widely recognised. It may indicate that the frameworks are operating from different assumptions about the context. Understanding the divergence is part of the work.

Alt-TAB is free. It is operated by Away from Keyboard (AFK) Inc. under licence from Xallarap, and it is free because the public interest case for accessible pre-deployment assessment is clear.

The organisations that use it are not all large enterprises with dedicated ethics teams. Many of them are small organisations, community groups, and public sector bodies that do not have the resources to commission a full assessment but do have the responsibility to make technology decisions carefully.

For those organisations, Alt-TAB provides a structured way to ask the questions that matter before a technology decision is made. It does not replace expert assessment for high-stakes decisions. But it provides a baseline that is grounded in 35 distinct analytical traditions, and that is a different kind of baseline than a single checklist or a single framework can provide.

The 35 frameworks are not a number chosen for effect. They are the number required to cover the territory adequately. That is the only reason the number is what it is.

Alt-TAB is available free at alttab.afk.org.au. For high-stakes technology decisions that require expert assessment, contact Xallarap.