“Mythos is a total weapon. You should need a weapons license to use it”

Anthropic and OpenAI are racing towards what is already known in Silicon Valley as the IPO (Initial Public Offering or initial public offering) of the artificial intelligence (IA), with valuations that are around 965,000 million and one trillion dollars respectively. In this scenario, the general director of Anthropic, Dario Amodei, has granted an interview to Bloomberg in which he reflects on his career in this industry and about the price that must be paid to maintain one’s values ​​when the business pushes in the opposite direction.

When asked about the decision to go for enterprise tools, such as Claude Code or Claude Cowork, instead of pursuing more popular consumer applications, Amodei was blunt: “Either you betray your values ​​or you become irrelevant.” And, according to this executive, choosing a business model that clashes head-on with a company’s values ​​ends up forcing that dilemma sooner or later.

However, Amodei’s warning is not limited to the business sphere. In this same interview he alerts us to a much more shocking and worrying phenomenon: the possibility that in the coming years skyrocketed GDP growth and very high unemployment will coexist, especially among entry-level office jobs.

The paradox that worries Amodei

The diagnosis that Anthropic’s CEO has been repeating for more than a year is the following: AI could eliminate half of entry-level office jobs within one to five years. Amodei insists that this headline is often taken out of context and that its real message has never been purely catastrophic. According to him, it is a call to anticipate with specific macroeconomic policies, such as taxes on the use of tokens or the redesign of templates together with companies.

The mechanism he describes is simple, although it is also disturbing. When 90% of a job is automated, the remaining 10% becomes much more productive thanks to that leverage. The problem comes when automation is approaching 100% and there is no task left to cling to. Within Anthropic itself, he maintains, they are already experiencing this transition: AI today writes practically all of the code, although for the moment it continues to make engineers more productive. It does not replace them completely.

When 90% of a job is automated, the remaining 10% becomes much more productive thanks to that leverage

This alert has earned him criticism from both sides. Jensen Huang, the CEO of Nvidia, has accused him of confusing tasks with jobs, while other voices speak directly of a “marketing of the apocalypse” that would benefit Anthropic. Amodei responds by recalling that in his essay “The adolescence of technology” He dedicated several pages to that same distinction and to a list of possible solutions, from private philanthropy to state intervention.

This is not the only front on which Amodei has chosen to put limits on its own technology. In the same interview he reveals that Anthropic retained Mythosits most advanced model for cybersecurity, because it is capable of autonomously going through the entire chain of a computer attack and finding exploitable vulnerabilities on its own. The companies that tested it in the early phase directly asked him to I won’t publish it. “Mythos is a total weapon. You should need a weapons license to use it,” he admits. Amodei acknowledges that this decision has cost him dearly in commercial terms, although he insists that publishing it before the rest of the network is patched would be much more dangerous.

Amodei himself frames this insistence within the same logic that he applied to his business model: he prefers to sound the alarm even if it is uncomfortable than to sweeten the diagnosis so as not to harm the valuation of his company on the way to its IPO. No one yet knows which side will win this race between productivity and unemployment. Amodei, at least, has already made clear which alternative he rejects.

Image | Wikimedia

More information | Bloomberg

In Xataka | China has a plan to win the AI ​​war against the US. And DeepSeek is its champion

In Xataka | Anthropic is one step away from being worth as much as Samsung. And what the market is buying is not Claude

Leave your vote

Leave a Comment

GIPHY App Key not set. Please check settings

Log In

Forgot password?

Forgot password?

Enter your account data and we will send you a link to reset your password.

Your password reset link appears to be invalid or expired.

Log in

Privacy Policy

Add to Collection

No Collections

Here you'll find all collections you've created before.