Generative AI Tools Expose Corporate Secrets Through User Prompts
A significant portion of employee interactions with generative AI tools is inadvertently leaking sensitive corporate data, posing serious security and compliance risks for organisations worldwide.
A comprehensive analysis by Harmonic Security, involving tens of thousands of prompts submitted to platforms such as ChatGPT, Copilot, Claude, Gemini, and Perplexity, revealed that 8.5% of these interactions contained sensitive information. Notably, 45.77% of the compromised data pertained to customer information, including billing details and authentication credentials. Employee-related data, such as payroll records and personal identifiers, constituted 26.68%, while legal and financial documents accounted for 14.95%. Security-related information, including access keys and internal protocols, made up 6.88%, and proprietary source code comprised 5.64% of the sensitive data identified.
The prevalence of free-tier usage among employees exacerbates the risk. In 2024, 63.8% of ChatGPT users operated on the free tier, with 53.5% of sensitive prompts entered through these accounts. Similar patterns were observed across other platforms, with 58.62% of Gemini users, 75% of Claude users, and 50.48% of Perplexity users utilizing free versions. These free tiers often lack robust security features, increasing the likelihood of data exposure.
ADVERTISEMENT
Anna Collard, Senior Vice President of Content Strategy & Evangelist at KnowBe4 Africa, highlighted the unintentional nature of these data leaks. She noted that users often underestimate the sensitivity of the information they input into AI platforms, leading to inadvertent disclosures. Collard emphasized that the casual and conversational nature of generative AI tools can lower users' guards, resulting in the sharing of confidential information that, when aggregated, can be exploited by malicious actors for targeted attacks.
The issue is compounded by the lack of comprehensive governance policies within organizations. A study by Dimensional Research and SailPoint found that while 96% of IT professionals acknowledge the security threats posed by autonomous AI agents, only 54% have full visibility into AI agent activities, and a mere 44% have established governance policies. Furthermore, 23% of IT professionals reported instances where AI agents were manipulated into revealing access credentials, and 80% observed unintended actions by these agents, such as accessing unauthorized systems or sharing inappropriate data.
The rapid adoption of generative AI tools, driven by their potential to enhance productivity and innovation, has outpaced the development of adequate security measures. Organizations are now grappling with the challenge of balancing the benefits of AI integration with the imperative to protect sensitive data. Experts advocate for the implementation of stringent oversight mechanisms, including robust access controls and comprehensive user education programs, to mitigate the risks associated with generative AI usage.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Campaign ME
10 hours ago
- Campaign ME
MiQ launches MiQ Sigma to unify programmatic ecosystem through AI
Global programmatic media partner MiQ today announced the launch of MiQ Sigma, a transformative AI-powered advertising technology trained on the world's most connected data set. Built with programmatic excellence at its core, MiQ Sigma unites over 300 diverse data feeds, spanning 700 trillion consumer signals across what consumers are watching on TV, browsing on the web, and buying in stores. Sophisticated AI is used across the platform to improve planning, audience development, and activation, allowing traders to drive market-leading outcomes for brands and agencies. MiQ Sigma is now the only advertising tool that connects the entire programmatic ecosystem in one place. This unified platform brings together data, technology, and campaign workflows, while applying advanced AI to make smarter decisions with speed. With the help of its trading agent, traders can reach valuable audiences across multiple DSP and SSP platforms and through different programmatically-enabled media across any screen. Wassim Mneimneh, MENAT Managing Director at MiQ, comments, 'MiQ Sigma represents a major step forward for MENAT advertisers. As the region accelerates its AI ambitions – including plans to build the world's largest AI campus outside the US in the UAE – it's clear it is a vital player in shaping the future of AI adoption. MiQ Sigma transforms the promise of AI into advertising performance, tackling one of the region's biggest challenges – fragmented data across digital channels.' In early testing, MiQ Sigma has shown to increase conversion rate by 132 per cent and reduce cost per action by 57 per cent. This is accomplished through key features, including: Sigma intelligence: Powerful visualisation featuring hundreds of diverse data feeds, spanning 'watching, browsing and buying' behaviours of over 1.7 billion global audience profiles Powerful visualisation featuring hundreds of diverse data feeds, spanning 'watching, browsing and buying' behaviours of over 1.7 billion global audience profiles Trading agent : Campaign management and optimisation features supported by an interactive trading agent, trained on 15 years of MiQ trading data and underpinned by the three leading LLMs (Claude, Gemini, and ChatGPT). Through simple, natural language commands, the agent enables traders to make quicker multi-DSP decisions and take immediate action : Campaign management and optimisation features supported by an interactive trading agent, trained on 15 years of MiQ trading data and underpinned by the three leading LLMs (Claude, Gemini, and ChatGPT). Through simple, natural language commands, the agent enables traders to make quicker multi-DSP decisions and take immediate action Gen AI personas: Custom audience profiles, built by generative-AI and accessed through natural language prompts, bring audience planning to life in seconds. MiQ Sigma is available immediately to MENAT outbound clients. For more information or to request a demo, visit:
Gulf Today
13 hours ago
- Gulf Today
Is AI sparking a cognitive revolution?
Artificial Intelligence began as a quest to simulate the human brain, according to the research by Clinical Professor Wolfgang Messner at the University of South Carolina. Is it now in the process of transforming the human brain's role in daily life? The Industrial Revolution diminished the need for manual labour. As someone who researches the application of AI in international business, I can't help but wonder whether it is spurring a cognitive revolution, obviating the need for certain cognitive processes as it reshapes how students, workers and artists write, design and decide. Graphic designers use AI to quickly create a slate of potential logos for their clients. Marketers test how AI-generated customer profiles will respond to ad campaigns. Software engineers deploy AI coding assistants. Students wield AI to draft essays in record time -- and teachers use similar tools to provide feedback. The economic and cultural implications are profound. What happens to the writer who no longer struggles with the perfect phrase, or the designer who no longer sketches dozens of variations before finding the right one? Will they become increasingly dependent on these cognitive prosthetics, similar to how using GPS diminishes navigation skills? And how can human creativity and critical thinking be preserved in an age of algorithmic abundance? We've been here before. The Industrial Revolution replaced artisanal craftsmanship with mechanized production, enabling goods to be replicated and manufactured on a mass scale. Shoes, cars and crops could be produced efficiently and uniformly. But products also became more bland, predictable and stripped of individuality. Craftsmanship retreated to the margins, as a luxury or a form of resistance. Today, there's a similar risk with the automation of thought. Generative AI tempts users to conflate speed with quality, productivity with originality. The danger is not that AI will fail us, but that people will accept the mediocrity of its outputs as the norm. When everything is fast, frictionless and "good enough," there's the risk of losing the depth, nuance and intellectual richness that define exceptional human work. Despite the name, AI doesn't actually think. Tools such as ChatGPT, Claude and Gemini process massive volumes of human-created content, often scraped from the internet without context or permission. Their outputs are statistical predictions of what word or pixel is likely to follow based on patterns in data they've processed. They are, in essence, mirrors that reflect collective human creative output back to users - rearranged and recombined, but fundamentally derivative. And this, in many ways, is precisely why they work so well. Consider the countless emails people write, the slide decks strategy consultants prepare and the advertisements that suffuse social media feeds. Much of this content follows predictable patterns and established formulas. It has been there before, in one form or the other. Generative AI excels at producing competent-sounding content - lists, summaries, press releases, advertisements - that bears the signs of human creation without that spark of ingenuity. It thrives in contexts where the demand for originality is low and when "good enough" is, well, good enough. Yet, even in a world of formulaic content, AI can be surprisingly helpful. In one set of experiments, researchers tasked people with completing various creative challenges. They found that those who used generative AI produced ideas that were, on average, more creative, outperforming participants who used web searches or no aids at all. In other words, AI can, in fact, elevate baseline creative performance. However, further analysis revealed a critical trade-off: Reliance on AI systems for brainstorming significantly reduced the diversity of ideas produced, which is a crucial element for creative breakthroughs. The systems tend to converge toward a predictable middle rather than exploring unconventional possibilities at the edges. I wasn't surprised by these findings. My students and I have found that the outputs of generative AI systems are most closely aligned with the values and worldviews of wealthy, English-speaking nations. This inherent bias quite naturally constrains the diversity of ideas these systems can generate. More troubling still, brief interactions with AI systems can subtly reshape how people approach problems and imagine solutions. One set of experiments tasked participants with making medical diagnoses with the help of AI. However, the researchers designed the experiment so that AI would give some participants flawed suggestions. Even after those participants stopped using the AI tool, they tended to unconsciously adopt those biases and make errors in their own decisions. What begins as a convenient shortcut risks becoming a self-reinforcing loop of diminishing originality - not because these tools produce objectively poor content, but because they quietly narrow the bandwidth of human creativity itself. True creativity, innovation and research are not just probabilistic recombinations of past data. They require conceptual leaps, cross-disciplinary thinking and real-world experience. These are qualities AI cannot replicate. It cannot invent the future. It can only remix the past. What AI generates may satisfy a short-term need: a quick summary, a plausible design, a passable script. But it rarely transforms, and genuine originality risks being drowned in a sea of algorithmic sameness. The challenge, then, isn't just technological. It's cultural. How can the irreplaceable value of human creativity be preserved amid this flood of synthetic content? The historical parallel with industrialisation offers both caution and hope. Mechanization displaced many workers but also gave rise to new forms of labor, education and prosperity. Similarly, while AI systems may automate some cognitive tasks, they may also open up new intellectual frontiers by simulating intellectual abilities. In doing so, they may take on creative responsibilities, such as inventing novel processes or developing criteria to evaluate their own outputs. This transformation is only at its early stages. Each new generation of AI models will produce outputs that once seemed like the purview of science fiction. The responsibility lies with professionals, educators and policymakers to shape this cognitive revolution with intention. Will it lead to intellectual flourishing or dependency? To a renaissance of human creativity or its gradual obsolescence? The answer, for now, is up in the air. Associated Press
Web Release
a day ago
- Web Release
NotebookLM Audio Overviews now available in Arabic
Google is launching Audio Overviews on NotebookLM in more than 50 languages including Arabic, starting this week. Audio Overviews , which turn uploaded sources into engaging, podcast-like conversations, were launched late last year in English. With Gemini's native audio support, people can now use Audio Overviews in their preferred language, and more updates will come to the feature based on users' feedback. Audio Overviews are generated in the preferred language of the user's account. This update also introduces a new 'Output Language' option in NotebookLM's settings. People can change the language at any time, making it easy to create multilingual content or study materials as needed. For example, a teacher preparing a lesson on – climate change in the region can share resources in various languages — like an Arabic documentary, an English research paper and Spanish study reports — with students. The students can upload these and can generate an Audio Overview of key insights in their preferred language. This capability breaks down language barriers and makes the information more accessible to everyone.
