AI rules are splitting between oversight of powerful models and rules for how organisations use AI in consequential decisions. In an analysis for Eurasia Review, Burak Oktenli argues that responsibility should remain visible as a system passes from model provider to integrator to the organisation using it.
Watch Desk analysis
What happened
Oktenli points to Colorado’s 26 October public hearing on proposed rules for automated decision-making technologies, ahead of a revised state law taking effect on 1 January 2027. He contrasts this focus on decisions affecting people with rules that scrutinise powerful models upstream, including the EU’s AI Act and California’s frontier-model law.
His proposed bridge is a short “consequential-use statement” for significant deployments. It would identify the decision a system may influence, the responsible institution, the roles of suppliers and integrators, relevant limits and evidence, and routes for review or redress. The Eurasia Review analysis presents this as a proposal, not an existing global requirement.
Why it matters
An AI model can be developed in one country, connected to local data and rules by an integrator elsewhere, then used by an employer, hospital or financial institution in a third jurisdiction. A report about the model alone may not reveal who configured the system, what decision it influenced or who can put a bad outcome right.
Oktenli argues that providers, integrators and deployers hold different pieces of the evidence and should not be able to pass responsibility down the chain. The analysis also distinguishes binding laws from voluntary guidance and government policy, a useful distinction in a debate where “AI rules” can otherwise sound like one large, interchangeable rulebook.
Our read
The handoff is where a lot of real-world accountability can go missing. A deployment-focused record could help make responsibility easier to trace without pretending every chatbot needs a new licence. But a form is only useful if people can challenge decisions and regulators have the power to act. Paperwork is not a remedy, however elegantly formatted.
What to watch
- Whether Colorado’s revised law leads to concrete rules for consequential AI use.
- Whether regulators or industry adopt deployment records that identify each party’s role.
- How requirements distinguish high-impact uses from ordinary, low-stakes tools.
Discussion spark: For consequential AI decisions, should the organisation using the system carry the main responsibility, or should model providers and integrators share it by law?
Sources and evidence
- Source update (13 November 2026, 02:15 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.