Wired report says Anthropic argues AI safety needs tight control

In short: Anthropic says advanced AI can only be made safe if the company keeps strong control over how it is built and used.

What's going on

Anthropic has spent years warning that very advanced AI could be dangerous. The company has pointed to risks like helping people cause large-scale harm and creating social instability. At the same time, it has become one of the major companies building and selling powerful AI models.

A new Wired report describes a growing tension around that message. Critics say Anthropic is quickly collecting a lot of influence by building the tools, setting many of the rules, and choosing who gets access. Anthropic’s response, according to the report, is that this concentration is part of responsible development.

Anthropic’s AI model is called Claude. AI models are the “brains” behind chatbots and other AI products, like a very advanced autocomplete that can write and answer questions. The report also notes that Anthropic has been courting large customers, including the US military.

Financially, Anthropic has also become much bigger. Wired reports the company was recently valued at almost $1 trillion, which signals how important investors think these tools could become.

What to watch

This debate matters because it is really about who should act as the gatekeeper for powerful AI, companies, governments, or some mix of both. If a few companies keep tight control, AI may be easier to limit in the short term, like keeping a dangerous tool locked up. But it also raises questions about oversight, competition, and what happens if the “lock” is controlled by the same group building the tool.

Source: Wired