Evening Scan

The day's biggest AI story wasn't a product launch — it was Anthropic getting caught, then backing down. OpenAI is reportedly preparing for a price war it expects Anthropic to start.

$409M — The fine South Korea just handed Coupang for a 2025 data breach that exposed roughly 33.7 million accounts, a record for the country's privacy watchdog. (Bloomberg)

Anthropic Backtracks on Fable 5's Secret Sabotage of AI Research

After backlash over covert restrictions buried in Claude 3.5's system card, Anthropic reversed course and promised the limitations would be visible. (Wired)

OpenAI Is Considering Drastic Token Price Cuts (paywall)

OpenAI is apparently so confident Anthropic will slash prices that it's moving first — competitive pressure doing what regulators haven't. (Wall Street Journal)

South Korea Fines Coupang $409M for Massive Data Breach (paywall)

The breach also escalated into a diplomatic conflict with the US. (Bloomberg)

Xbox Warns of a "Reset" in the Next 100 Days

New CEO Asha Sharma and CCO Matt Booty laid out the problem plainly: annual revenue is down roughly $500M over the past five years, and the reset is coming regardless of fanbase readiness. (The Verge)