What it is: Smart Reports became available to all Claude Enterprise customers on October 28 after a beta. Claude analyzes a sample of a team’s conversations over up to 28 days and summarizes what they worked on, how often, what it cost and where work got stuck.

The controversial part: Reports count how many sessions were “personal or off-topic,” how many employees were involved and what those sessions cost. It’s the first Claude feature where the AI reads employees’ conversations and summarizes them for management.

Built-in safeguards: Names are hidden by default, personal sessions are only shown as totals, and sensitive topics like health and union activity are blocked from admin drill-downs. Anthropic says these protections can’t be turned off.

Anthropic’s warning: The company says reports are for understanding adoption and planning AI investment, not for judging individual performance or making decisions about staff.

Availability: It’s off by default and must be turned on by an Enterprise owner. Each organization gets 10 free reports a month. It’s not available on Pro, Max or Team plans, or for some high-security setups like HIPAA.

The plan: On Thursday’s earnings call, CEO Brian Niccol said Starbucks will explore drive-thru experiences that recognize participating customers as they arrive, show them personalized digital menus and make payment easier.

Still early: Niccol called the idea early-stage and said Starbucks will “test, learn, and scale what works.” There’s no timeline, and Starbucks hasn’t explained how customer recognition would work.

AI for baristas: Separately, starting in fiscal 2027, Starbucks plans to test background AI at drive-thrus to help baristas enter orders more accurately and quickly. Customers won’t see it.

More AI coming: Starbucks plans to add AI to search and recommendations in its app and may eventually allow conversational ordering. Customers can already order through ChatGPT, and baristas use an AI tool called Green Dot Assist for recipe questions.

The competition: Taco Bell is testing menu boards that change car by car, and McDonald’s is piloting an AI feature that offers returning customers their usual order.

Teen weapons limits: ChatGPT will now restrict practical information about weapons and security systems for users under 18, even when there’s no sign of violent intent. Its rules already banned helping anyone plan violence or helping teens get weapons.

For all users: ChatGPT can automatically end chats where someone seems fixated on violence, and OpenAI may temporarily pause accounts after several such chats. It’s also limiting details about past mass attacks, including the weapons used and how they were obtained.

Parents and age checks: OpenAI improved its age prediction to move suspected under-18 users into the teen experience, and plans to notify parents if their teen’s account is paused for repeatedly discussing violence.

Transparency: OpenAI referred more than 200 cases of potential violent threats to law enforcement in Q3 2026. It will now publish a Community Safety Report on violence prevention every six months.

Why now: Earlier this month, Common Sense Media rated ChatGPT for Teens an “Unacceptable Risk,” finding it often alerted parents late or not at all about conversations on serious topics. The new safeguards roll out over the next few weeks.

Secret Key Let your family unlock your passwords if you can't

Godwit An open-source, AI-native email client

SpaceO Give AI agents their own screen on your Mac

Arsaze Cursor for video editing. Make one cut, Tab the rest

Local Commander Give your computers to any AI, even from your phone