Signal President Warns: AI Chatbots Are a Privacy Risk – Here’s How to Protect Yourself

Meredith Whittaker, the president of the encrypted messaging app Signal, recently warned that AI chatbots pose a growing threat to user privacy. Her comments, reported by SC Media, come as millions of people turn to tools like ChatGPT, Bing Chat, and Google Bard for everyday tasks. For the average user, the convenience of these chatbots can hide serious data risks.

What Happened

In a statement covered by SC Media, Whittaker pointed out that most commercial AI chatbots operate on a fundamentally different privacy model than apps like Signal. While Signal uses end-to-end encryption to ensure that neither the company nor anyone else can read your messages, many chatbot providers store your conversations on their servers, often in plain text or with weak protection.

Whittaker also noted that some chatbot platforms share user data with third parties for training, moderation, or advertising. She described this as a “black box” where users have little visibility into what happens to their inputs. Signal itself does not incorporate AI chatbots, reinforcing its position that privacy and convenience should not be traded against each other.

Why It Matters

The warning is timely. AI chatbots are becoming embedded in workplaces, schools, and homes. People use them to draft emails, get medical advice, or even discuss personal problems. What many don’t realize:

  • Chats can be stored indefinitely. Even if you delete a conversation, the provider may keep copies for months.
  • No end-to-end encryption. Unlike a secure messaging app, most chatbots can see and log everything you type.
  • Inputs can be used for training. Several companies have faced backlash for using customer conversations to improve their models without explicit consent.
  • Data breaches expose raw chats. Because many chatbots store user inputs in databases, a breach can leak sensitive information you thought was private.

Even a seemingly innocent question – like asking for help planning a surprise party – could reveal your address, phone number, or travel dates. And once that data is out there, you can’t take it back.

What Readers Can Do

You don’t have to stop using AI chatbots altogether. But you can take concrete steps to reduce your exposure:

  1. Avoid sharing sensitive personal information. Think before you paste a confidential email, a medical record, or your home address. If it’s something you wouldn’t want a stranger to read, don’t type it into a chatbot.

  2. Use incognito or temporary chat modes. Some platforms now offer a “no-save” option. For example, ChatGPT’s incognito mode prevents conversations from being used for training. Turn this on if available.

  3. Check your account settings. Look for data retention and sharing policies. On many services, you can opt out of having your chats used for training. This doesn’t guarantee deletion, but it limits secondary use.

  4. Don’t log in with Google, Facebook, or Apple ID. Using a single sign‑on method often ties your chatbot activity to your main account, making it easier for companies to link and sell your data. Create a separate account with a minimal profile.

  5. Prefer privacy‑focused alternatives. Consider open‑source chatbots that run locally on your device, such as GPT4All or Llama.cpp. These keep your data offline. For cloud‑based options, look for services that advertise zero‑data retention and independent audits.

  6. Clear your history regularly. If the platform allows deletion, make it a habit. But remember that deleted does not mean gone – logs may persist on backup servers.

The goal is not paranoia but informed caution. As Whittaker’s warning makes clear, the default privacy settings on most chatbots are weak. Adjusting them takes only a few minutes but can significantly reduce risk.

Sources

  • SC Media: “Signal president warns about AI chatbot privacy risks” (June 22, 2026)
  • PBS: “AP report: Hegseth warns Anthropic…” (February 24, 2026) – contextual background on AI and national security
  • SC Media: “The ROME Incident: When the AI agent becomes the insider threat” (March 10, 2026) – examples of AI‑related data risks

This article is based on publicly reported statements and widely known privacy practices. Specific implementation details for each chatbot may vary; consult the provider’s privacy policy for up‑to‑date information.