The Mexican business daily Expansion reported on 11 August that the research group Mexico, como vamos? had matched the first-quarter 2026 round of the Encuesta Nacional de Ocupacion y Empleo (ENOE), Mexico's national occupation and employment survey, against the generative-AI exposure index the International Labour Organization built with NASK, Poland's state research institute. On that account, 2.93 million Mexican workers, 4.9% of employment, hold occupations in the highest-exposure band, and 1.87 million of them are women.1
The distribution matters more than the total. Informality runs at 66% among workers in minimally exposed occupations and at 16.5% in the most exposed ones, so the jobs a model can reach in Mexico are disproportionately the jobs that come with a contract, a payroll record and social security. The street vendor and the building labourer are, on this reading, further from automation than the analyst in an office tower.
That inverts the intuition most of this beat runs on, and it changes what a Mexican policy response would have to protect. Formal employment is the destination the country's labour reforms have spent a decade pushing people towards. It is also the segment where a young graduate expects a first job, which is why the study frames the risk as a threat to the first formal post rather than to the workforce at large.
The evidence sits one step back from the primary document. The report's own page returned an HTTP 403 to repeated fetch attempts, so every figure here rests on the outlet's account of the study rather than on its methodology notes. Index-matching studies of this kind have multiplied since Stanford put its AI Economic Indicators dashboard into the open in June, and they inherit whatever the underlying exposure score assumes. The ENOE half of the calculation is a national statistical product; the exposure half is a model.
