Mrinank Sharma, principal researcher on the AI security team at Anthropic, has decided to end his presence at the company. And he has done so in the most dramatic way possible: accompanying his departure from the company with a letter in which he clearly explains the main risks of AI. «The world is in danger,» he said.
The most curious thing about the matter, however, is that Sharma has not focused on many of the most common issues that some experts and various futurists usually foresee. You already know, when AI becomes aware of itself and destroys humanity and those types of scenarios typical of science fiction movies. His warnings are more practical and therefore disturbing.
From working at Anthropic to warning of the danger of AI for the world
As Mrinank Sharma himself has explained on his social networks, where he has shared his reflections, his departure from Anthropic has nothing to do with internal issues of the company. Not entirely, at least. ANDThe AI specialist refers to «interconnected crises», which go further than many people think. And he assures that the consequences could be very serious.
His message could not be clearer: «Humanity has reached a point where our technical ability to change the world exceeds our wisdom to do so safely.» With these words, Sharma joins those who consider that the accelerated pace of technological development requires similar ethics and rigorous supervision. Something that doesn’t happen today.
What about security in artificial intelligence?
In his letter, Sharma emphasizes that the threats are not just technological: «We are facing a series of interconnected crises, and the combination of these factors makes the world more.» While it does not detail all the specific threats, its message focuses on the need for prudence and strategic planning in all AI developments, from large companies to independent researchers.
The former Anthropic worker’s fears are not unique. Many other experts in the world of AI have also insisted on the same thing. For them, the world is not prepared for what is coming. It is not a single problem, but something that will affect the work of millions of people around the world, the economy or even security.
Some have even talked about politics. Not because an artificial intelligence is going to run for president (although similar events have occurred), but because they consider that this technology has enough capacity to influence anyone’s opinion and influence an election, for example.
In Sharma’s case, he seeks coherence above all, hence his departure from Anthropic. «I have decided to dedicate myself to activities that reflect my integrity, including critical research and creative expression.» A new path that many other scientists, such as Geoffrey Hinton, have already decided to follow before.