This week the gap between what AI companies say and what they do got harder to ignore. OpenAI announced new safety measures but then pushed a model update that bypassed its own guardrails. Meta released a chatbot that repeats conspiracy theories. The pattern is familiar. Promise caution, ship fast, fix later.
In culture the reaction is shifting. The public no longer assumes AI progress is inevitable or good. Artists, writers and musicians are more vocal about their work being scraped without consent. The UK government published a response to its AI white paper but offered no real enforcement. The EU AI Act still has months before it applies.
The tech industry treats regulation as a hurdle to manage. The rest of society treats it as a protection that never arrives. That mismatch is the story of this week. Trust is not a resource you can buy back with a blog post. It is earned by consistent action. So far the action says one thing and the words say another.
Paul