It's not often that companies that stand to gain from a technology call for new laws that might constrain them. But Smith is worried enough about the spread of surveillance systems with powerful facial recognition that he's calling for lawmakers to act now.
Tech companies are faced with a "commercial race to the bottom", which should have a "floor of responsibility" that allows competition but outlaws the use of facial recognition in ways that harm democratic freedom or enable discrimination.
Smith is concerned that unchecked facial recognition will increase the risk of biased decisions and outcomes, and may invade people's privacy, while its use for mass surveillance could harm democratic freedoms.
He argues that facial-recognition laws should require tech companies to provide transparent documentation that explains the capabilities and limitations of their facial-recognition tech.
The laws should also require providers of facial-recognition services to undergo third-party testing to check for accuracy and unfair bias.
"While we're hopeful that market forces may eventually solve issues relating to bias and discrimination, we've witnessed an increasing risk of facial-recognition services being used in ways that may adversely affect consumers and citizens -- today," writes Smith.
The legislation should also force organizations that use facial recognition to review its impact and ensure that using the technology isn't an escape route for complying with anti-discrimination laws.
Other areas that should be covered include clearly notifying consumers where facial recognition is in use, and require consumers to give consent to the use of facial recognition when entering premises.
Microsoft also wants constraints on law enforcement use of facial recognition when monitoring people of interest in public places.
Smith argues this tactic should only be allowed with a court order, or in emergency, such as the risk of death or serious injury to a person.