Why this matters

AI assistants are safe and genuinely helpful — but they're also new online services run by companies. A few simple habits let you get all the benefit without the avoidable risks.

The core habits

• Never paste true secrets. Passwords, full card numbers, government IDs, or other people's private data don't belong in a chat box. If you'd hesitate to email it to a stranger, don't paste it.
• Assume conversations may be reviewed. On most free consumer plans, your chats can be sampled by the company to improve future models. You can often turn this off in settings, or use a business plan where your data isn't used for training.
• Verify high-stakes advice. For anything medical, legal, or financial, treat the answer as a starting point and confirm it with a qualified human or a primary source.
• Be honest about AI use. Where it counts — schoolwork, professional deliverables — disclose it or check the rules. Passing off AI work as entirely your own can have real consequences.

Spotting the newer risks

As AI spreads, so do AI-powered scams. Two to keep on your radar:

• Deepfakes and voice clones. A familiar voice or face can now be faked convincingly. If a "relative" calls in a panic asking for money, hang up and verify through a number you already trust.
• Too-perfect messages. Phishing emails used to be full of typos; AI makes them fluent. Judge a message by what it asks you to do — click, pay, share a code — not by how polished it reads.

A simple mental model

Treat AI as a helpful assistant, not a vault and not an authority. Share with it what you'd share with a smart coworker you just met — and double-check anything important before you act on it.

Try this: Open your assistant's privacy or data settings right now and find the toggle for "improve the model" or "training." Decide consciously whether you want it on or off. Thirty seconds, and you're in control.