ChatGPT and privacy
Geplaatst op: 5 December 2023
Privacy and AI
Privacy is one of the most important human rights. It gives the right to individuals to protect their personal life and information from unlawful interference by others, including the government. This right includes various aspects such as the right to protection of personal data, the right to control personal information, and the right to personal autonomy and freedom. However, due to increasing data collection by AI and the growing power of emerging technologies, individuals in both the public and private sectors will increasingly be tracked and monitored, often without enough anonymity or consent.
ChatGPT unethically processes your data
ChatGPT uses large amounts of text data collected from various sources, including public websites and social media While the data is anonymized, there is a chance that sensitive information such as names, locations, and personal opinions could be incorporated. Yes, you read that right, private information made publicly available. As cybersecurity experts, it’s therefore our job to make users aware of the privacy risks associated with the use of AI models such as ChatGPT. But that’s not all… But that’s not all…
Discrimination and racism: AI systems like ChatGPT are filled with biases.
Al machine learning models, including AI models that perform specific tasks, are trained based on datasets, i.e. mountains of data and information on which the model bases its output. It is the only way to get the “intelligence” of an AI as close as possible to that of humans, at least that’s what the researchers of the AI system claim. ChatGPT, for example, is trained with more than 300 billion words, that’s about 570 GB of data. That’s exactly where the problem lies.
When you harvest large, unsorted data sets from the internet, it is inevitable that they’ll contain some form of biased information. This information then influences the models. Even though researchers do use filters to prevent AI-models like ChatGPT from providing false information after collecting data, these filters are not 100% waterproof. This can cause harmful biases, or worse: ChatGPT can generate answers that are plainly racist or discriminatory.
It is also important to point out that the data on which ChatGPT generates responses is outdated. In short, the data does not reflect where society currently stands in its development and progress. And, perhaps the most important: Even the researchers and developers themselves are often biased. This is because the world of tech simply is an extremely homogeneous sector dominated by, yes, none other than ‘white men’.
If we continue to ignore the privacy risks and ethical issues when using ChatGPT, among others, we are contributing to a greater invasion of our own privacy and injustice. It’s important that users take their responsibility and protect their own privacy. In doing so, they should also be aware of the ethical implications when using ChatGPT.
How to protect yourself
If you still want to use ChatGPT, it is essential to acknowledge the limitations of ChatGPT andto take responsibility for the contents that are generated and analysed under the influence of this type of technology. In addition, there are two other important measures that you can take yourself to better guard your privacy:
- 1. Avoid sharing sensitive information with ChatGPT: When getting started in ChatGPT, we recommend that you only share non-sensitive information in the prompts and do not provide any personal information. Do not share any pictures or videos, and certainly no contact details.
- 2. Do not use ChatGPT for processing sensitive information: It is also wise to disregard ChatGPT when you need to process sensitive information, like responding to emails or processing personal data. This will prevent the AI from using your private information for learning purposes and or sharing it with other users.
Remain aware of the privacy risks and take the necessary measures to protect your privacy. Do you want to know more about how to protect your data online? One of our experts will be happy to help you on your way!
5 useful tips for dealing with data management and data security
Geplaatst op: 28 May 2022
The tasks of municipalities have expanded considerably in recent years. From implementing and safeguarding Social Support tasks to the Environment Act and the social domain. Consequently, they are responsible for hundreds, if not thousands, of different systems that are all interconnected. Systems where piles of sensitive information are stored and analyzed. It’s therefore extremely important that the storage and processing of vulnerable data is done properly. In this article, we identify five trends that contribute to data security in municipalities.
1. Map security risks
An easily preventable security risk in municipalities and government agencies is the fragmentation of information or systems. They often do not have an overview of where the potential security risks or weaknesses are in the various systems. Adding to that, the division of roles among staff is also often a sensitive point that municipalities encounter. Make sure it’s clear which employees are authorised to view and process sensitive information. Data security starts with identifying the risks.
2. Employee awareness and accountability
An accident is often just around the corner. Human error is the most common cause of a data breach. It is therefore essential that all employees within the organisation are aware of the hazards and risks. This includes the accountability they take on when working with sensitive data. By training employees, you create a strong foundation for overall data security in the organisation.
3. Invest in new technologies
More and more municipalities are opting for a digital transformation. Technologies and data-driven practices such as machine learning, data analytics and block chain offer municipalities numerous opportunities to make work more efficient as well as safer. Block chain technology can, for example, help municipalities ensure quality care in Social Support tasks and transparent communication between different care providers. Investing in digital transformation with an eye on data privacy and data management will benefit many municipalities. Take advantage of it!
More tips?
Download the PDF for free, using the form below. In it, we provide additional tips that are important to keep cyber security strong and in order.
The largest cyber attack in history
Geplaatst op: 2 May 2022
In the last few years, cyber attacks have become much more common. We also often talk about it in our blogs. But text and explanation can sometimes be a bit bland and don’t speak to the imagination. That’s why in this blog we’ll give you an example of one of the biggest cyber attacks in recent history.
WannaCry (2017)
In 2017, one of the largest ransomware attacks ever took place. The attack is known as the WannaCry-attack because the ransomware used is named WannaCry. In 2017, the NSA (National Security Agency) created a tool that exploited a security vulnerability in Microsoft Software. Obviously, this tool was not made to distribute ransomware, but the NSA itself suffered a cyber-attack in which the so-called ETERNALBLUE tool was stolen and published online, making it publicly available.
Solution
Microsoft had been informed about the vulnerability a month before the leak and soon had a so-called patch (solution) that resolved the vulnerability in the security of the software. However, and you’ll probably recognize this in your own organization, not everyone had downloaded the update with the patch (“remind me later”) It soon became clear that many people were still using outdated computers and/or software. Hackers launched a ransomware attack using the ETERNALBLUE tool and infected 200,000 computers across 150 countries within a day.
Within a few hours a cyber security expert (who was also an ex-hacker) found a so called ‘kill switch’, a kind of self-destruct button in the software that caused the ransomware to shut down and stop spreading itself.
Effects
Ultimately, it’s estimated that this attack cost between 100 million and several billion dollars in damage. The WannaCry cyber attack received attention partly because of its impact on the NHS (National Health Service). The WannaCry attack disabled about 70,000 computers in hospitals all across England. The cyber attack was carried out on a scale never seen before and could have been much worse, according to experts. Using the ETERNALBLUE tool was a very well thought-out action, according to experts, but other parts of the attack including the ”kill switch” looked like amateur mistakes. Without these mistakes and the clever performance of the ex-hacker and cyber security expert Marcus Hutchins, this cyber attack could have been a lot worse with disastrous consequences.
An important lesson
Perhaps the most important lesson that we can take out of this story is: Always check if your software is up to date! Always. Updates are extensively tested in advance by developers and in all cases serve as an improvement to the current software package. This could be an update to the operation or interface, but in most cases it is a security update. Updating your software prevents you from falling victim to a (un)known security vulnerability.