Stop Asking Them Nicely

We have spent years asking technology companies to behave responsibly. Please protect our data. Please pay creators. Please reduce harms. Please make your systems transparent. Please consider workers. Please use less energy. Please do not deploy something dangerous. Please regulate yourselves before governments feel compelled to do it for you. And then we produce principles, voluntary commitments, industry standards, ethics guidelines and another round of Responsible AI frameworks explaining how seriously everybody takes all of this.

There is a basic conflict hidden underneath that politeness. We are asking profit-seeking institutions to voluntarily choose social interests when those interests collide with their own. Sometimes they will. Companies are not incapable of responsible behavior, and the people working inside them are not cartoon villains. But expecting voluntary restraint to consistently win against competition, investor pressure, market capture and billions in sunk costs is not governance. It is hope dressed up as a framework.

At some point politics has to rediscover a fairly basic idea: concentrated power is not controlled by asking it to be nicer. We regulate, tax, prohibit, mandate, inspect, redistribute, create public alternatives and give workers and citizens actual rights precisely because interests conflict. Responsible AI cannot mean that society writes a wish list and waits for corporations to voluntarily comply. If something matters enough that we cannot afford for companies to ignore it, then perhaps we should stop asking.