AI Safety Jobs Are So Competitive, Even Experts Can't Get Hired

There's a common belief that the field of AI safety desperately needs more talented people. But according to one insider, that's simply not true. In fact, there are far more qualified people trying to get into AI safety than there are jobs available. For example, a recent Anthropic Fellows program reportedly received over 10,000 applicants. Even someone who had worked at top AI safety organizations like Redwood and METR was rejected. The competition is fierce, and the acceptance rates are tiny—around 5% for some programs.

So why does this matter to you? Because you might be one of the many people who hear the call to 'pivot into AI safety' and think it's a promising career move. But the reality is that these jobs are extremely competitive and high-paying—some salaries start at over $400,000. They're not struggling to attract talent; they're struggling to handle the flood of applicants. Telling people to drop everything and switch is actually harmful. It sends them into a grueling application process with little chance of success, when they could be contributing in other valuable ways.

The author also worries that this myth of being 'talent-constrained' is holding the field back. If AI safety organizations believe they just need to wait for the right genius to come along, they might ignore fixable problems within their own teams and processes. A regular startup that treated its talent pipeline this way would fail quickly. The message is clear: if you're considering a pivot to AI safety, do your research first. The field may not be as welcoming as it seems, and your skills might be better used elsewhere.