The meteoric fall of Nvidia's Jensen Huang, following controversy over his personal relationships and US national security, has raised questions about judgement, trust and the people leading the world's most powerful technology companies.
But it also prompts a broader question. How much do we really know about the people behind the technologies we increasingly trust with our information? And how private is that information we’re sending in confidence kept?
I know I’m not alone in treating artificial intelligence as my personal doctor, giving it detailed information about my health in exchange for bespoke medical advice without needing a co-pay.
It’s also my instant therapist, answering personal questions about how I should engage with problematic people in my life. Basically information I probably wouldn’t want anyone to know.
We use ChatGPT, Google's Gemini, and Anthropic's Claude to research, analyse documents, draft correspondence, and solve problems.
In addition, people may share sensitive personal information, professional dilemmas and confidential business material.
Sometimes, all we do is ask a question. And that is revealing in itself.
But what actually happens to that information once we press enter? Who can see it, and where does it go?
The answer may surprise people who assume their conversations with AI are entirely private.
How much is out there
When we enter information into a cloud-based AI platform, it travels to computing infrastructure operated by the provider or its partners. There it is processed, and depending on the service and privacy settings, it may be stored, checked by automated systems, reviewed by authorised humans or used to improve AI models.
These companies are surprisingly candid about this in their posts.
OpenAI says that authorised employees and contracted service providers may access consumer conversations for safety investigations, technical support, legal matters and model improvement, where permitted.
Google's Gemini documentation goes further. It confirms that trained human reviewers, including external service providers, may examine information processed from connected applications. This can include excerpts and summaries of emails and documents, and sometimes the complete contents of a short email or file.
Google advises users not to connect applications containing confidential information they would not want a reviewer to see.
And if this is said publicly, imagine what else is happening behind closed doors that should give us pause.
It's not a free-for-all, as these companies have access restrictions and operate under national and international laws. But the key message is clear: just because your entered information is not public, it doesn’t mean it's private.
You might assume that, because you are using a personal AI account, it's a closed-loop conversation. But that depends on how you access it.
Using office Wi-Fi alone does not ordinarily allow your employer to read encrypted AI conversations, although it may reveal which services you visit. A company laptop equipped with monitoring software, however, could expose more. Microsoft's own security tools, for example, can inspect information typed or pasted into AI prompts.
The situation is different again when you use an employer-managed AI account.
OpenAI confirms that authorised administrators may, depending on permissions and configuration, access or export employees' prompts, files and conversation histories.
Anthropic similarly confirms that designated owners of Claude for Work accounts can request exports containing employee conversations.
In other words, what feels like a private conversation may be accessible to your employer, depending on the account, device and security systems you use.
Rest assured, encryption generally prevents employers from reading AI prompts over office Wi-Fi, according to the Electronic Frontier Foundation, a US-based digital rights organisation. This is true unless additional monitoring tools are in place.
Nevertheless, the combination of workplace monitoring, managed accounts and third-party AI platforms creates a complicated privacy landscape that few employees fully understand.
There is another side to this. What happens when executives handling sensitive information use personal AI accounts to analyse confidential acquisitions, strategic plans or financial projections?
Research from Cyberhaven's 2026 AI Adoption and Risk Report found that 39.7 per cent of observed data movements into AI tools involved sensitive corporate information.
Meanwhile, IBM's 2025 breach research found that one in five organisations studied had experienced breaches linked to unauthorised AI use.
The risks are no longer theoretical.
The recent $6.34 billion Binance hack highlights this danger. In its November investigation, the cryptocurrency exchange revealed that attackers had exploited compromised employee devices and an internal AI tool as part of the breach.
Our privacy rights?
Beyond corporate security lies a more fundamental question for everyday people: What rights do we have over the information we voluntarily give these systems?
The UAE's Personal Data Protection Law, under Articles 13 and 15, and Europe's General Data Protection Regulation, under Articles 15 to 17, give individuals rights to access information about their personal data, correct inaccuracies and, under certain conditions, request its deletion.
But these rights are not absolute. Having the legal right to request deletion does not necessarily mean every trace of an AI conversation can be erased from all systems.
Google states that certain conversations reviewed by humans may remain stored for up to three years, even after the user deletes their activity. Switching off Gemini's activity history still allows temporary retention for operational and safety purposes.
We can choose privacy settings, but those choices do not always amount to complete control.
For the UAE, where AI is central to economic ambition and business transformation, this deserves particular attention. Companies need clear policies for AI use, while individuals need to understand the implications of using personal and corporate accounts.
The issue is not whether we should abandon AI. Its benefits are enormous, world-changing even.
The question is whether we can trust these technologies and the organisations behind them to keep our private information actually private, on our own terms.
History clearly shows that regulation often lags behind technological innovation, leaving society exposed to risks that take years to address.
We saw this with the internet and social media platforms designed to exploit reward mechanisms, particularly among young people. More than two decades later, countries such as Australia and the UAE are now introducing tougher age restrictions to protect children. Are we repeating the same mistake with AI, allowing its rapid adoption to outpace safeguards? Must we wait another two decades to act?
We scrutinise what technology leaders do with their private lives. Perhaps we should be equally curious about what their companies do with ours.

