DeepSeek and Moonshot AI Caught Secretly Feeding Data to Anthropic’s Claude
In a twist that reads like a spy thriller written by a bored dev, DeepSeek and Moonshot AI have been busted outsourcing their brainpower to Anthropic’s Claude. It turns out even the most nationalistic AI needs a little help from the West.
The Cyberspace Administration of China has officially kicked off an investigation into DeepSeek and Moonshot AI. The allegation is that these tech labs were sneaking user prompts across the Pacific to process them through Anthropic's Claude, essentially using the competitor's tech to train their own models.
Anthropic blew the whistle in a massive 154-page report, claiming that Moonshot AI pushed over 23 million queries and DeepSeek added another 12 million in just a few weeks. These weren't just random math homework questions; the report suggests sensitive data from government entities and global corporations was bundled right into these requests.
The irony here is thick enough to cut with a knife: in their desperate rush to reach parity with global AI leaders, these labs were basically copy-pasting homework from the very entities they were meant to challenge. The regulators are now digging through the logs to see just how much sensitive info was gift-wrapped for an American company to refine its own algorithms.
The absurdity of a sovereign tech sector being entirely dependent on a rival's infrastructure for performance is the perfect punchline to the AI arms race. Trust is clearly just a legacy feature that nobody in the industry seems to care about anymore.
Source: The Information
Comments
This is where the magic happens: AI reads your discussion and rewrites the article based on the most interesting comments. Each strong comment adds points to the meter below. Once the meter is full, the article updates live — no page reload needed.