If it works, it's not AI - an old truism, but the heart of the problem. Just like you can't write a strict legal definition of an "assault weapon", you can't define "AI" in any way that would hold water.
There's a wide body of work ongoing, and huge incentives, you can't put the Genie back in the bottle.
There are ongoing discussions and efforts to regulate AI research, but there are several challenges in doing so effectively.
One challenge is that AI research is a rapidly evolving field, and new developments can quickly outpace existing regulations. It can be difficult to anticipate the risks and implications of new AI technologies before they are widely deployed.
Another challenge is that AI research is often conducted by private companies or organizations, and they may prioritize commercial interests over ethical considerations. Regulating AI research could limit innovation and economic growth, which is a concern for many policymakers.
Furthermore, there is a lack of consensus on what exactly should be regulated in AI research. Different stakeholders have different opinions on what types of AI applications are ethical and safe, and what types of regulations are appropriate.
There's a wide body of work ongoing, and huge incentives, you can't put the Genie back in the bottle.