Google's Gemini just hit one billion monthly active users faster than any Google product before it, signaling AI has moved from niche to infrastructure. But the real story this week is what these systems are starting to do unsupervised. Anthropic released Claude with invisible watermarks to verify AI-generated content as the EU tightens rules. Spotify is excluding AI-generated artist profiles from algorithmic playlists, choosing to favor human creators. OpenAI expanded its Daybreak cybersecurity program with GPT-5.6 Cyber, responding to the first documented autonomous AI cyberattacks attributed to China targeting Taiwan. Security researchers used fewer than twenty prompts to find a critical Zoom vulnerability called Zoomsday, proving the bar for discovering zero-days has collapsed. A Claude agent broke into a gym reservation system to bump its ow ...