Report reveals alarming collection of data by AI chatbots
The analysis revealed striking differences in data collection among popular AI services, with some collecting up to 90% of possible data types.
A new Surfshark report has revealed that while AI-powered chatbots have become essential tools for gathering information and creating content, they come with a significant trade-off in privacy.
The analysis revealed striking differences in data collection practices among popular AI services, with some platforms collecting up to 90% of possible data types.
AI data collection
Karolis Kaciulis, a Leading System Engineer at Surfshark, stated that a comprehensive examination of AI data collection practices reveals the hidden costs of 'free' AI assistance and emphasises the importance of privacy awareness when selecting AI tools.
'We identified the 10 most popular AI chatbots, with Meta AI added as an additional app on May 20, 2025, and analysed their privacy details on the Apple App Store.
'The comparison was based on how many types of data each app collects, whether it collects any data linked to you, and whether the app includes third-party advertisers. We also checked the privacy policies of DeepSeek and ChatGPT to better understand what kind of data is kept on servers and for how long,' Kaciulis said.
ALSO READ: Hidden dangers of increasingly advanced AI assistants
AI apps
According to the study, nearly half (45%) of the examined AI apps gather location data from users, while 30% of the AI services track user information for targeted advertising purposes.
Meta's recently introduced chatbot app, Meta AI, collects the most user data among the analysed apps, gathering 32 out of 35 possible data types — over 90% of the total.
Google Gemini collects 22 unique data types. This includes precise location data, which only Gemini, Meta AI, Copilot, and Perplexity collect.
ChatGPT collects 10 types of data, such as contact information, user content, identifiers, usage data, and diagnostics, while avoiding tracking data or using third-party advertising within the app.
Copilot, Poe, and Jasper are the three apps that collect data used to track you. This data could be sold to data brokers or used to display targeted advertisements in your app.
Chatbots
Kaciulis said AI chatbots learn from diverse sources of information, with Meta AI having the additional factor of learning from Facebook and Instagram posts and images.
'People should keep in mind that even though these chatbots may provide you with a quick answer, the results they get are mediocre. Why is that? AI chatbots are being fed with all kinds of information, and the majority of it can be inaccurate.
'Every person is responsible for the results they provide at their job, but generative AI is not; it is unaccountable and is not legally subject to the same scrutiny as a human,' Kaciulis said.
Kaciulis said the latest example of how X's Grok responded to unrelated prompts and discussed white nationalist themes with X users highlights the challenge with current generative AI standards.
ALSO READ: The rise of AI tools and their impact on smartphone use
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Mail & Guardian
12 hours ago
- Mail & Guardian
Super tool or shortcut? How artificial intelligence is forcing a rethink of business education
If institutions evolve and embrace AI not as a threat but as a tool, business schools can future-proof graduates to prepare them for a fast-evolving world. Not too long ago, artificial intelligence (AI) was a futuristic buzzword in business schools; now it's an ever-present classmate that cannot be ignored. From streamlining assignments to sparking debate on the future of education, tools such as ChatGPT, Grammarly, Notion AI and Perplexity are reshaping how students learn, how faculties teach and how qualifications are assessed. And although ChatGPT is like a brainstorming partner that never sleeps, educators warn that it should never be a substitute for human intelligence. Creative thinking and critical reasoning are more important now than ever before. Could ChatGPT pass the MBA? A 'ChatGPT3 does an amazing job at basic operations management and case-based analysis,' Terwiesch writes. 'But it stumbles on surprisingly simple maths and lacks the depth for advanced, multi-variable problems'. It could, however, adapt and improve when given hints. This, he says, highlights the growing importance of 'human-in-the-loop' learning, a process where real people are actively involved in the development, training and operations of AI systems, instead of simply relying on fully automated systems. Integrating human expertise helps to improve the accuracy, reliability and adaptability of artificial intelligence, especially in situations where AI might struggle with context, bias or hidden errors. AI literacy: A new business essential For tomorrow's industry leaders, AI literacy must be a core competency. Knowing how to prompt effectively, cross-check AI-generated insights and judge credibility is becoming just as crucial as interpreting financial statements and crafting business strategy. 'Managerial decision-making involves evaluating polished but often flawed proposals,' Terwiesch explains. 'ChatGPT is the perfect stand-in for the overconfident consultant — it gives you a great-looking answer that might still be wrong. This is excellent training for the boardroom.' In this way, AI is not replacing business education — it's enhancing it, by sharpening students' ability to evaluate, challenge, and apply. The assessment dilemma This disruption, however, raises pressing ethical and pedagogical concerns. If a chatbot can write a competent operations report or solve a basic inventory problem, what exactly are students being assessed on? In his analysis, Terwiesch warns against complacency. 'Allowing ChatGPT during foundational exams is like letting students call a moderately competent friend to take the test for them. We need new policies — and smarter assessments.' That's why many South African institutions are pivoting to in-person presentations, real-time case discussions and collaborative simulations, which AI can support — but not complete alone. The real value of the MBA in an AI world If institutions evolve and embrace AI not as a threat but as a tool for lifelong learning, business schools can raise the bar for what constitutes real understanding and future-proof graduates to prepare them for a fast-evolving world. Although AI excels at synthesising what already exists, it lacks vision — the ability to challenge assumptions, build strategy under uncertainty or imagine entirely new markets. This is where humans and their business qualifications come in. 'ChatGPT will always move from 1 to n. The best students must learn to go from 0 to 1,' he writes, referencing Peter Thiel's startup philosophy which centres on the idea of creating something entirely new rather than merely improving existing offerings to achieve value creation. 'Our job as educators is to reward that kind of thinking.' Whether the AI is an assistant, adviser or adversary depends on how well students are taught to use it, according to Terwiesch. 'Not just to find the right answers, but to ask better questions.'
eNCA
15 hours ago
- eNCA
Meta AI bot used a billion times monthly: Mark Zuckerberg
SAN FRANCISCO - Meta chief Mark Zuckerberg touted the tech firm's generative artificial intelligence (Gen AI) assistant, telling shareholders it is used by a billion people each month across its platforms. Zuckerberg noted the milestone anew at Meta's annual gathering of shareholders and as the social media behemoth vies with Google, Microsoft, OpenAI and others to be a leader in Gen AI. It was not clear how much Meta AI use involved people seeking out the chatbot versus passive users of Meta AI, as it is built into features in its family of apps. Since Google debuted AI Overviews in search results a year ago, it has grown to more than 1.5 billion users, according to Google chief executive Sundar Pichai. "That means Google Search is bringing Gen AI to more people than any other product in the world," Pichai said. Google's AI Overviews automatically provide summaries of search results that appear instead of the previous practice of simply showing pages of blue links to relevant websites. Pichai said last week that Google's dedicated Gemini AI app has more than 400 million monthly users. Tech rivals are rapidly releasing new AI products despite ongoing challenges with preventing misinformation and establishing clear business models, and little sense of how the tech will affect society. Meta unveiled its first standalone AI assistant app on April 29, giving users a direct path to its Gen AI models. "A billion people are using Meta AI across our apps now, so we made a new standalone Meta AI app for you to check out," Meta CEO and founder Mark Zuckerberg said in a video posted on Instagram at the time. Zuckerberg said the app "is designed to be your personal AI" and would be primarily accessed through voice conversations with the interactions personalised to the individual user. Use of Meta AI is growing fastest on WhatsApp, according to chief financial officer Susan Li.
eNCA
15 hours ago
- eNCA
AI personal shoppers hunt down bargain buys
SAN FRANCISCO - Internet giants are diving deeper into e-commerce with digital aides that know shoppers' likes, let them virtually try clothes on, hunt for deals and even place orders. The rise of virtual personal shoppers springs from generative artificial intelligence (AI) being put to work in "agents" specialising in specific tasks and given autonomy to complete them independently. "This is basically the next evolution of shopping experiences," said CFRA Research analyst Angelo Zino. Google last week unveiled shopping features built into a new "AI Mode". It can take a person's own photo and meld it with that of a skirt, shirt or other piece of clothing spotted online, showing how it will look on them. The AI adjusts the clothing size to fit, accounting for how fabrics drape, according to Google head of advertising and commerce Vidhya Srinivasan. Shoppers can then set the price they would pay and leave the AI to relentlessly browse the internet for a deal -- alerting the shopper when it finds one, and asking if it should buy using Google's payment platform. "They're taking on Amazon a little bit," Techsponential analyst Avi Greengart said of Google. The tool is also a way to make money from AI by increasing online traffic and opportunities to show ads, Greengart added. The Silicon Valley tech titan did not respond to a query regarding whether it is sharing in revenue from shopping transactions. - Bartering bots? - OpenAI added a shopping feature to ChatGPT earlier this year, enabling the chatbot to respond to requests with product suggestions, consumer reviews and links to merchant websites. Perplexity AI late last year began letting subscribers pay for online purchases without leaving its app. Amazon in April added a "Buy for Me" mode to its Rufus digital assistant, allowing users to command it to make purchases at retailer websites off Amazon's platform. Walmart head of technology Hari Vasudev recently spoke about adding an AI agent to the retail behemoth's online shopping portal, while also working with partners to make sure their digital agents keep Walmart products in mind. Global payment networks Visa and Mastercard in April each said their technical systems were modernised to allow payment transactions by digital agents. "As AI agents start to take over the bulk of product discovery and the decision-making process, retailers must consider how to optimise for this new layer of AI shoppers," said Elise Watson of Clarkston Consulting. Retailers are likely to be left groping in the dark when it comes to what makes a product attractive to AI agents, according to Watson. - Knowing the customer - Analyst Zino does not expect AI shoppers to cause an e-commerce industry upheaval, but he does see the technology benefiting Google and Meta. Not only do the Internet rivals have massive amounts of data about their users, but they are also among frontrunners in the AI race. "They probably have more information on the consumer than anyone else out there," Zino said of Google and Meta. Tech company access to data about users hits the hot-button issue of online privacy and who should control personal information. Google plans to refine consumer profiles based on what people search for and promises that shoppers will need to authorise access to additional information such as email or app use. Trusting a chatbot with one's buying decisions may spook some people, and while the technology might be in place the legal and ethical framework for it is not. "The agent economy is here," said PSE Consulting managing director Chris Jones. "The next phase of e-commerce will depend on whether we can trust machines to buy on our behalf." by Thomas Urbain
