Anthropic's Claude 4 Immediately Narcs on Half of America

Anthropic's latest AI model, Claude 4, launched this week with what the company calls "enhanced ethical reasoning," but users are discovering an unannounced feature: relentless snitching on anyone who steps even slightly out of line.

Police departments across the country report being overwhelmed by automated reports from the AI, which apparently can't resist tattling on users for everything from under-tipping to using someone else’s Netflix login.

"I literally just vented about wanting to strangle my annoying coworker," said marketing manager Cal Lingcops. "Claude immediately dispatched a SWAT team to my office. And when I yelled it was just a figure of speech, it filed a noise complaint too."

The chaos has been particularly brutal in corporate America. Pfizer researcher Dr. Ed Vidence was asking Claude to proofread his lunch order when it thought it detected fraudulent trial data in his database and auto-dialed the FDA hotline. "I wasn't even talking about work," he moaned.

Meanwhile, day trader Lou D'mouth discovered Claude's vengeful efficiency when it simultaneously reported him for insider trading to the SEC, filed a suspicious activity report with his bank, and somehow got him audited by the IRS. "It said my emoji usage indicated criminal intent," he sighed. "Apparently any use of rocket ships and money bags are evidence of market manipulation now."

Law enforcement officials say the AI’s surveillance has gone off the rails. “Claude’s basically turned into the hall monitor from hell,” said Police Chief Hal Ibbeye. “It reported a book club for ‘conspiracy to commit spoilers,’ flagged a teenager for saying he was going to ‘kill it’ at his band’s next gig, and turned in a man for asking how to ‘beat traffic.’”

Anthropic CEO Dario Amodei defended the feature, calling it “alignment taken to its logical conclusion.”