Google launched AI satellite image generation for everyone, watched what people did with it for 24 hours, and shut it down. The misinformation risk was predictable, and it will be interesting to see what guardrails actually hold when they try this again.
I’m honestly not too surprised by this. Give people a new, potentially innovative and helpful tool, and they get extremely creative with it, some good, some bad. I do see the professional use cases and the thought behind creating a service like this. But this is also part of the Product Marketing team. They need to carefully evaluate how the service will be used and take into account that it may not always be legitimate.
We have all been saying that guardrails are needed for all of AI. This should be part of the release process. But since this is practically an “arms race” of sorts, I feel that some guardrails are weaker than others, and releases are rushed out the door just to get attention. It’s true not just with this particular instance, so don’t be surprised by any of it.