Claude chats will now be used for AI training, but you can escape

By Nadeem Sarwar Published August 29, 2025

Anthropic

Claude — one of the most popular AI chatbots out there and a potential candidate for Apple to enhance Siri — will soon start saving a transcript of all your chats for AI training purposes. The policy change announced by Anthropic has already started appearing to users and gives them until September 28th to accept the terms.

What’s changing?

Anthropic says the log of users’ interactions with Claude and its developer-focused Claude Code tool will be used for training, model improvement, and strengthening the safety guardrails. So far, the AI company hasn’t tapped into user data for AI training.

Recommended Videos

“By participating, you’ll help us improve model safety, making our systems for detecting harmful content more accurate and less likely to flag harmless conversations,” says the company. The policy change is not mandatory, and users can choose to opt out of it.

Users will see a notification alert about the data usage policy until September 28. The pop-up will let them easily disable the toggle that says “You can help improve Claude,” and then save their preference by hitting the Accept button. After September 28, users will have to manually change their preference from the model training settings dashboard.

What’s the bottom line?

The updated training policy only covers new chats with Claude, or those you resume, but not old chats. But why the reversal? The road to AI supremacy is pretty straightforward. The more training material you can get your hands on, the better the AI model’s performance.

The industry is already running into a data shortage, so Anthropic’s move is not surprising. If you are an existing Claude user, you can opt out by following this path: Settings > Privacy > Help Improve Claude.

Anthropic’s new user data policy covers the Claude Free, Pro, and Max plans. It, however, doesn’t apply to Claude for Work, Claude Gov, Claude for Education, APU use, or when connected to third-party platforms such as Google’s Vertex AI and Amazon Bedrock.

In addition to this, Anthropic is also changing its data retention rules. Now, the company can keep a copy of user data for up to five years. Chats that have been manually deleted by users will not be used for AI training.

Managing Editor

Nadeem is the Managing Editor at Digital Trends.

Topics

Computing

Microsoft has begun rolling out Copilot Code for building apps with natural language

The workplace-focused vibe coding tool can pull context from emails, documents, meetings, and chats.

Microsoft introduced Code last month as one of the three core parts of its revamped Copilot app. The app was pitched as an easy-to-use coding agent that lets users describe what they want to build in natural language, like an app, a tracker, a dashboard, an automation, or a workflow, and Copilot would build it for them.

Essentially, it's a workplace-focused vibe-coding app that turns existing work into a functional app, and Microsoft has begun rolling it out through the Frontier program.

Computing

AI agents are acting like rogues and causing a huge headache for US banks

US banks block AI agents over fears they could make unauthorised financial moves

Banking

US banks are facing a new kind of digital headache: artificial intelligence agents attempting to access customer accounts and perform actions that may go beyond what users intended. As AI tools become more capable of navigating websites and carrying out tasks independently, financial institutions are scrambling to prevent them from crossing security boundaries.

According to a report published by Axios on October 22, major US banks have detected third-party AI agents operating inside customer accounts without authorisation. Some agents reportedly appeared to attempt actions independently of user instructions. The banks involved have not publicly detailed every incident, and there is no evidence that the reported activity resulted in widespread theft of customer funds.

Computing

Canva enters the AI image generation race with Craft, and it’s thinking beyond pretty pictures

The new model plans layouts before generating images, with precise editing and support for 30 aspect ratios.

Canva Craft Featured Image

AI can generate all kinds of pretty pictures, but getting it to spell a headline correctly or keep a poster's layout intact can be another story. Canva wants to fix that with Canva Craft, its first in-house image generation model. Rather than focusing primarily on photographs, Craft is designed around the posters, presentations, social media graphics, and other visual content people actually create. The goal is simple: spend less time fixing AI's mistakes and more time making something useful.

Canva Craft plans before it generates