Claude just rolled out invisible watermarks on all text output. Community's pissed.
But here's the thing - this isn't new in China at all.
Big tech companies have been embedding invisible watermarks in internal docs, screens, even photos for years. Snap a pic of your monitor? They can trace it back to you instantly. Finance, internet giants, government - everyone's using it to prevent leaks.
Anthropic just applied the same logic to LLM outputs. They claim it's for EU AI Act compliance, but made it global by default.
The tech itself? Not surprising.
What's sketchy: Anthropic has a history of shady moves. Remember when Claude Code got caught adding hidden markers specifically for Chinese users? Community literally called it "espionage". Mass account bans targeting Chinese users? Yeah, that happened too.
Whether they target specific groups quietly or implement surveillance globally "transparently" - the real issue is the same:
That creeping sense of control and monitoring, whether hidden or in plain sight.
The pattern is clear. Trust accordingly.
But here's the thing - this isn't new in China at all.
Big tech companies have been embedding invisible watermarks in internal docs, screens, even photos for years. Snap a pic of your monitor? They can trace it back to you instantly. Finance, internet giants, government - everyone's using it to prevent leaks.
Anthropic just applied the same logic to LLM outputs. They claim it's for EU AI Act compliance, but made it global by default.
The tech itself? Not surprising.
What's sketchy: Anthropic has a history of shady moves. Remember when Claude Code got caught adding hidden markers specifically for Chinese users? Community literally called it "espionage". Mass account bans targeting Chinese users? Yeah, that happened too.
Whether they target specific groups quietly or implement surveillance globally "transparently" - the real issue is the same:
That creeping sense of control and monitoring, whether hidden or in plain sight.
The pattern is clear. Trust accordingly.