
WhatsApp privacy alert: Gemini AI can access message previews even when turned off. What to do now?
On its official support page, Google explained that Gemini Apps provide 'direct access to Google AI,' and that interactions with Gemini are temporarily stored—up to 72 hours—even if the user has opted out of activity tracking. This means that even with Gemini activity turned off, data from messaging apps like WhatsApp may still be processed and stored briefly.
What Does Gemini Actually Do With Your Messages?
While Gemini's expanded access is intended to help users perform tasks like sending messages or setting reminders via voice commands or Assistant features, the lack of clarity around data usage is what has triggered concern. Some privacy advocates and users believe this could compromise personal information.
According to Google, Gemini does not train its AI on your data, and interactions are stored only temporarily to ensure the service's functionality and safety. However, as highlighted by Neowin, even when not actively reading your chats, Gemini may still access and respond to WhatsApp messages through Android's notification system or utility functions—particularly when used in combination with Google Assistant.
This level of access may include reading message previews, images, and potentially sending replies on your behalf.
How to Limit or Disable Gemini Access
If users are uncomfortable with this level of integration, Google does provide options—though limited—for managing Gemini's activity:
Turn Off Gemini Apps Activity
Open the Gemini app.
Tap the profile icon on the top-right corner.
Select 'Gemini Apps Activity.'
Use the toggle to disable the feature.
Limit App-Level Access
In the Gemini app, tap your profile picture again.
Select 'Apps.'
Choose which apps Gemini is allowed to connect with.
Fully Disable Gemini
You can also disable the Gemini app altogether if you want to prevent any interaction or tracking.
It's important to note that even after turning off the activity feature, Google will still retain your data for up to 72 hours for what it claims is safety and security purposes.
Broader AI Integration Raises Red Flags
The concerns surrounding Gemini echo similar controversies involving other tech giants. Earlier this year, Microsoft faced backlash over its Recall feature in Windows 11, which captured screenshots of user activity, including messages. Only apps like Signal were able to prevent this tracking.
Meta also faced criticism when its AI tool was embedded into WhatsApp chats, although users can now block that integration. The common thread across these developments is a growing unease over AI systems having near-unrestricted access to private data without explicit or informed user consent.

Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles


Indian Express
9 minutes ago
- Indian Express
Microsoft Copilot's new ‘Smart' mode may be powered by OpenAI's GPT-5
ChatGPT maker OpenAI is gearing up to unveil GPT-5, its upcoming large language model. Rumour has it that the much-anticipated AI model will be launched sometime next week and bring significant improvements in areas like advanced reasoning and autonomous task execution. Ahead of GPT-5's launch, Microsoft Copilot is now reportedly rolling out a new 'Smart' mode for select users that may be powered by OpenAI's upcoming large language model. First spotted by TestingCatalog, the new Smart mode is available for select users on both the Copilot app and website and can be accessed by clicking on the model selector that appears just below the text field. According to Microsoft, the new Smart mode enables Copilot to 'think deeply or quickly based on the task.' Since when is 'Smart Mode' a thing in Copilot?? — Bughunter Nerd 🌎🌍🌏☮️❤️ (@bughunternerd) July 30, 2025 Copilot's much-rumoured GPT-5-powered Smart mode will change its response depending on the user query, meaning it will automatically switch between available large language models, so users don't have to. Microsoft's description of Copilot's Smart mode also suggests that GPT-5 will have both reasoning and non-reasoning capabilities, which is something OpenAI CEO Sam Altman has hinted at in the last few months. The upcoming AI model has been under development for some time, with several reports claiming that the company has pushed back its launch date, citing safety reasons. GPT-5 is expected to roll out sometime in August and come with different intelligence modes, but it is still unclear which level of intelligence Copilot will use. Microsoft has been shipping Copilot with all new Windows-powered devices, and while it does not enjoy the same level of popularity as ChatGPT, the tech giant wants to make it the easiest generative AI assistant one can use. And while reports have hinted that executives at Sam Altman's company were weighing accusations of anticompetitive practices against the tech giant, Microsoft's partnership with ChatGPT maker OpenAI has only grown stronger over the past few years,

The Hindu
9 minutes ago
- The Hindu
Google to sign EU's AI code of practice despite concerns
Alphabet's Google will sign the European Union's code of practice which aims to help companies comply with the bloc's landmark artificial intelligence rules, its global affairs president said in a blog post on Wednesday, though he voiced some concerns. The voluntary code of practice, drawn up by 13 independent experts, aims to provide legal certainty to signatories on how to meet requirements under the Artificial Intelligence Act (AI Act), such as issuing summaries of the content used to train their general-purpose AI models and complying with EU copyright law. "We do so with the hope that this code, as applied, will promote European citizens' and businesses' access to secure, first-rate AI tools as they become available," Kent Walker, who is also Alphabet's chief legal officer, said in the blog post. He added, however, that Google was concerned that the AI Act and code of practice risk slowing Europe's development and deployment of AI. "In particular, departures from EU copyright law, steps that slow approvals, or requirements that expose trade secrets could chill European model development and deployment, harming Europe's competitiveness," Walker said. Microsoft will likely sign the code, its president, Brad Smith, told Reuters earlier this month, while Meta Platforms declined to do so and cited the legal uncertainties for model developers.

Mint
9 minutes ago
- Mint
ChatGPT maker OpenAI has doubled revenues to $12 billion, generating $1 billion a month, says report
Sam Altman-led artificial intelligence leader OpenAI has close to doubled its revenues in the first seven months of 2025 (H1-2025), reaching an annualised revenue of around $12 billion, according to a report by The Information, citing sources. OpenAI did not respond to queries on the same, Reuters reported. The $12 billion number implies that OpenAI is generating $1 billion a month, The Information noted. The company has around 700 million weekly active users for its ChatGPT products used by both consumers and business customers, it added. The Microsoft-backed company has increased its cash burn projection to roughly $8 billion in 2025, up $1 billion from the cash burn it projected earlier in the year, the Information said. The firm has been lining up investors for the second $30 billion portion of its funding round, the report said, adding that shareholders Sequoia Capital and Tiger Global Management are investing hundreds of millions of dollars in the round. Investors, besides Japan's SoftBank , are close to finalizing $7.5 billion in commitments to that second portion of funding, the report said. The Japanese conglomerate's total agreed investment in OpenAI stood at $32 billion since first investing in Autumn 2024. (With inputs from Reuters)