Microsoft steps up calls for regulation of facial recognition tech; research group AI Now stresses urgency for companies to open their AI algorithms to auditing
AFTER A HELLISH year of tech scandals, even government-averse executives have started professing their openness to legislation.
Context & Ripple Effects
This is the second act of a campaign Microsoft began in July 2018, when President Brad Smith first asked governments to regulate facial recognition and defended the company's ICE contract as not involving the technology — a statement that immediately drew an ACLU moratorium call on government use of the tech. The December follow-up pairs Microsoft's renewed lobbying with AI Now's parallel demand that companies submit their algorithms to independent auditing, widening the debate from one product category to AI systems generally.
The arc matters because it worked: by mid-2020 Microsoft and Amazon were jointly pushing for federal regulation while local laws accumulated on their own, and Smith's DC playbook — refined across this fight — resurfaces in his later effort to shape AI regulation through proposals like a dedicated agency. The story is less about one technology than about a vendor deciding rules are safer than an unregulated free-for-all.
First-order effects
- Microsoft's lobbying raises the compliance bar for its own facial recognition deployments and for government customers, who now face both regulatory pressure and civil-society scrutiny — the ACLU's moratorium demand being the sharpest edge of it.
- AI Now's auditability demand puts every company selling AI systems on notice that voluntary self-assessment is no longer the accepted standard among policy groups.
Second-order effects
- Rivals Amazon and Google are pulled into the same regulatory frame — by 2020 they are co-lobbying with Microsoft rather than waiting for rules to be imposed, converting a competitive liability into a shared cost of doing business.
- An independent-audit market becomes a plausible growth area: if AI Now's framing holds, vendors need third-party assessors the way cloud providers eventually needed certifications, shifting some compliance spend from internal teams to auditors.
Third-order effects
- The pattern — a large vendor inviting regulation to shape it — becomes the template for how the industry approaches AI governance broadly, culminating in Smith's later advocacy for a dedicated AI regulator; whether regulators accept vendor-drafted frameworks or impose stricter terms is the genuine open question.
- If auditing becomes a condition of selling AI to governments, procurement itself turns into a governance lever, structurally favoring large vendors who can afford audits over smaller entrants.
The trend: Major AI vendors are shifting from resisting regulation to actively courting it, with facial recognition serving as the test case for who writes the rules.