OpenAI's new ChatGPT feature records keystrokes in plain text
openai privacy
| Source: Mastodon | Original article
OpenAI has added a ChatGPT feature that records users' keystrokes and saves them as plain‑text files, raising privacy concerns.
OpenAI has rolled out a new ChatGPT capability that records every keystroke a user makes and saves the data as plain‑text files. The feature, announced on the company’s blog and reported by The Next Web, captures raw input before it reaches the model, meaning that even incomplete sentences or accidental key presses are stored without encryption.
The move has sparked immediate privacy concerns. Storing keystrokes in an unprotected format creates a low‑cost target for malicious actors and raises questions about how long the data will be retained, who can access it, and whether it will be used for training or analytics. For users accustomed to end‑to‑end encryption in many messaging and productivity tools, the lack of safeguards feels like a step backward in data protection.
The development adds another layer to the scrutiny OpenAI has faced this year. As we reported on “OpenAI’s Unraveling Has Begun” on 20 August 2026, the company is already under pressure from regulators and privacy advocates over its data‑handling practices. This latest feature could accelerate calls for clearer transparency, stricter consent mechanisms, or even regulatory intervention in the EU and beyond.
What to watch next: OpenAI’s official response, including any updates to its privacy policy or security measures; reactions from consumer‑rights groups and potential investigations by data‑protection authorities; and whether the company will offer opt‑out options or encrypted storage for keystroke logs. The episode underscores the broader tension between AI convenience and user privacy that is reshaping the industry.
Sources
Back to AIPULSEN