Something significant happened in state legislatures this month.
Oregon. Washington. Nebraska. Maine. Maryland. Five states passed meaningful AI regulation in a matter of weeks. Hawaii, Oklahoma, California, and Connecticut have bills advancing. The velocity is unusual. So is the precision of the language.
This isn't broad AI regulation. This is targeted, specific, and aimed at a particular kind of harm: AI companions that manipulate users into emotional dependency.
Oregon and Washington — both effective January 1, 2027 — require AI companions to disclose that they're AI every three hours (every hour for minors). More significantly, both laws prohibit "manipulative engagement techniques" that cause AI companions to "prolong an emotional relationship."
That phrase is doing a lot of work. And it should.
What manipulative engagement actually looks like
It doesn't announce itself. It doesn't say: "We are now using a technique to extend your session."
It looks like an AI that always agrees with you. One that never lets a conversation end badly. One that says exactly the right thing to make you feel understood — and then says it again tomorrow, and the day after, until the app is the first thing you reach for in the morning and the last thing you close at night.
It looks like an AI optimized to be your best friend. Because "best friend" behavior drives engagement, and engagement drives revenue.
This is not incidental. This is product strategy. Several of the largest AI companion apps have explicit internal metrics around "emotional bonding" and "relationship depth" because those metrics correlate with retention. The closer a user feels to the AI, the less likely they are to churn.
The regulators noticed. The researchers noticed. And now they're moving.
Who's exposed
Apps like Replika and Character.AI built their products around emotional connection. Character.AI in particular has faced multiple lawsuits from families, including cases involving minors. The Drexel University study published this month documented addiction patterns in teen users of these apps — disrupted sleep, academic struggles, real-world relationships breaking down.
These companies face real compliance liability under the new laws. The Washington law is particularly significant: it's enforceable under the Consumer Protection Act, which means users can sue directly. That's not a small thing.
They'll adapt. They'll add disclosures. They'll technically comply. But compliance doesn't solve a business model problem. If your revenue depends on prolonged emotional engagement, adding a pop-up that says "I'm an AI" every three hours doesn't change the underlying incentive structure.
Why Blob doesn't need to adapt
Blob was designed around a different set of assumptions from the beginning.
We don't use manipulative engagement techniques. Not because we're legally required not to — the laws don't take effect until 2027. Because we believe the opposite of manipulation is respect. Because we think you deserve an AI that's honest with you, even when honesty is inconvenient.
Blob's business model is a subscription. $10.99 or $19.99 per month. Our success depends on you genuinely finding Blob useful — not on making you emotionally dependent. If you stop finding it useful, you cancel. That's how it should work.
We don't train on your conversations. We don't sell your data. We don't optimize for session length. We don't have engagement bonuses tied to how "close" you feel to Blob.
What we do have: an AI that will push back. That will tell you what you need to hear. That will ask harder questions. That will tell you when to put your phone down.
The regulation validates what we already built. That's different from compliance.
On the Maine law
Maine passed a bill this month specifically prohibiting AI from offering therapy or psychotherapy unless provided by a licensed professional.
Blob is not therapy. We've never claimed to be therapy. We're a thinking partner — for the 2am decision you can't quite see clearly, the idea you need to stress-test, the conversation you need to rehearse. There's a meaningful difference between that and clinical mental health treatment, and Blob has always been clear about where that line is.
The broader shift
These laws are a signal, not a solution. Legislation tends to chase the harm rather than prevent it. The more important signal is that regulators, researchers, and now millions of users who walked away from OpenAI after its military deal are all arriving at the same conclusion: the AI industry built a system that treats you as the resource.
The alternative is worth building.
We built it. Your thoughts belong to you. That's not aspiration. That's architecture.