Anthropic says it blocked misuse attempts targeting Claude

In short: Anthropic says it disrupted attempts to misuse its Claude AI model, including campaigns linked to Russia and China, and it described other high risk requests it has seen.

What happened

Anthropic, the company behind the AI model Claude, said it has been working to stop people from using Claude for harmful purposes. An AI model is a computer program that can write and answer questions, a bit like a very fast assistant.

Recent reporting says Anthropic disrupted campaigns connected to Russia and China that were targeting its systems. The company also shared examples of attempted misuse, including requests that touched on biological weapons. It did not present this as a normal use case, but as something it is trying to prevent.

This comes during a week when several outlets also pointed to wider pressure on the AI industry. One theme is the cost of the hardware and electricity needed to run big AI systems, similar to how a growing city needs more power plants and roads. Another theme is government attention, including rules that may ask companies to show officials previews of very powerful models before they are released.

Why it matters

AI tools are now used in everyday work and school, so how companies handle misuse affects regular people. If safeguards are weak, the same tool that helps write a report could also help someone plan harm. The flip side is that stronger safeguards can limit what the tool will do, which can frustrate some users, so companies and governments are still trying to find the right balance.

Source: NYTimes