IBM Walks Away From Facial Recognition
IBM's CEO told Congress the company will stop selling general-purpose facial recognition software, citing surveillance and racial profiling risks.
Today IBM CEO Arvind Krishna sent a letter to Congress announcing that IBM will no longer offer general-purpose facial recognition or analysis software. Full stop. The company is walking away from a product category it has been developing for years, and it’s doing so publicly, in a letter tied directly to policy debate on Capitol Hill.
The letter is unusually blunt for a corporate policy statement. Krishna wrote that IBM firmly opposes the use of facial recognition technology for mass surveillance, racial profiling, and violations of basic human rights and freedoms — and that the company no longer offers this software for those purposes. Reading between the lines, that’s not a hedge. That’s IBM saying the technology, as currently deployed, is too easily bent toward exactly the abuses everyone worries about, and they’d rather not be in that business at all.
The timing isn’t subtle. This comes in the middle of nationwide protests over police brutality following George Floyd’s death, with police reform legislation actively being drafted in Congress. Facial recognition has been a flashpoint in that conversation for a while now — civil liberties groups and researchers have spent years flagging that these systems are demonstrably worse at identifying people with darker skin, and that police departments have quietly bought and deployed the tech with little public oversight or independent auditing. IBM pulling out reads like a company deciding the reputational and ethical liability isn’t worth whatever revenue this line was bringing in.
The pressure this puts on everyone else
IBM was never the biggest player here — Amazon’s Rekognition and Microsoft’s facial recognition offerings get far more attention, and far more scrutiny, from researchers and advocacy groups. Both companies have sold or pitched facial recognition tools to police departments and law enforcement agencies. Neither has announced anything close to what IBM just did.
That’s what makes this move interesting beyond IBM’s own business. It’s a shot across the bow. Now there’s a major, credible enterprise tech company on record saying this technology shouldn’t be sold for law enforcement surveillance use, at a moment when Congress is actually paying attention and protestors are in the streets demanding police accountability. That’s going to make it harder for Amazon and Microsoft to stay quiet. Employees at both companies have pushed back on facial recognition contracts before; expect that internal pressure to get louder this week, and expect reporters to be asking Jeff Bezos and Satya Nadella’s PR teams for a response very soon.
Worth noting: this doesn’t kill facial recognition as a technology. IBM can still walk this back, competitors can fill the gap, and plenty of smaller vendors with less to lose reputation-wise are already selling into this exact market. But when a company with IBM’s history in enterprise AI says no thanks to a lucrative product category, on ethical grounds, in a letter to Congress, it changes the shape of the argument. It’s no longer just civil liberties groups versus tech companies. It’s tech companies disagreeing with each other about where the line is.