OpenAI speaks about the past to ignore the present

Why tech companies suddenly feel compassionate about history.

OpenAI recently published a statement recognizing the importance of understanding slavery and historical injustice. It was beautifully written. Empathetic. They understand the importance of diversity, of recognition, of awareness.

At the same time, they train their AI models on billions of images without permission, payment, or dividend to those whose images they use. Disproportionately many from non-white people. Disproportionately many from poor countries. The origin of each datapoint is unknown. Compensation is zero.

This isn't a direct comparison. Slavery was horror. But the pattern is identical. Raw material is extracted from people who have no voice, is processed without their approval, and generates value they don't share.

Historical awareness without contemporary accountability is at worst only marketing. It says: we understand what we did wrong, so you can trust we're doing it right now. But meanwhile we're doing exactly the same thing, just more invisibly.

This is the danger of corporate wokeism. It feels like progress. It feels like recognition. But it's a distraction maneuver. While OpenAI speaks about the importance of understanding history, they rebuild the same power structure anew, only with algorithms instead of chains.

I say this not because I'm angry at OpenAI specifically. I say it because it's systemic. Tech companies take what isn't theirs, generate billions in value, and then say "we recognize historical injustice." That recognition without action is fake.

Real awareness would mean: we don't train on data whose sources we don't know. We pay the makers of that data. We ask permission. We share value.

In iRecord I try a different approach. Identity is yours, not mine. Biometric data stays local. SHA-256 hashing means you authenticate yourself without exposing yourself. Connection without Capture.

But this requires political will. Tech companies don't want this pattern to break. The pattern is what gives them power.

OpenAI's speech about slavery was probably genuine. It probably felt good to write. But it does nothing. It says everything. And saying while you do is the modern form of hypocrisy.

Sources: OpenAI 'Building Equitable AI' statement (2024); Robert Hartley, 'A History of Slavery and Anti-Slavery' (Oxford University Press, 2007); Timnit Gebru research on dataset bias and AI ethics