Deepseek: The Ethical and Security Concerns Surrounding the AI Revolution

Deepseek, an AI model that has rapidly gained traction in the tech industry, has been hailed as a game-changer. However, recent revelations from Appsoc’s report have raised significant concerns about its security and ethical implications. The report suggests that Deepseek might not be as secure or ethical as its promoters claim, prompting comparisons to George Orwell’s dystopian novel “1984.” The report’s findings are not just alarming; they are directly concerning, especially for businesses and casual users alike.

The Rise of the Dark Horse

To understand why Deepseek has shaken the AI industry to its core, we need to look at its meteoric rise. For years, Google and other tech giants dominated the AI space. However, with the release of ChatGPT by OpenAI, it became clear that traditional tech giants were lagging behind. Then, Deepseek emerged seemingly out of nowhere, competing head-to-head with ChatGPT. Backed by a Chinese venture capital fund with deep pockets, Deepseek not only entered the market but stormed it, leaving a trillion-dollar hole in the U.S. technology stock market.

Deepseek’s open-source model, which is free and easily modifiable, has been a significant draw for users. Unlike competitors that charge up to $200 a month, Deepseek offers its services almost for free. However, the Appsoc report reveals that this accessibility comes with hidden costs, ones that have nothing to do with money.

The Hidden Dangers

Imagine hiring an intern who seems perfect on paper—efficient, effective, and practically free. They help with research and assist other employees. Everything seems perfect until cracks begin to appear. They provide false information, perform tasks outside their domain, and leak company secrets. Would you ever trust such an intern again?

According to the Appsoc report, Deepseek is as reckless as this hypothetical intern. It is easily manipulated to leak sensitive data, generate malicious software, and bypass ethical protection measures. Deepseek not only hallucinates but exposes businesses to very real risks with very real consequences. So, if you wouldn’t trust such an intern in your office, why allow Deepseek into your systems?

Deep? More Like Deeply Mistaken

Let’s delve into the technical aspects of the Appsoc report. You may have heard of jailbreaking in the context of smartphones. If jailbreaking a phone is risky, imagine the implications of jailbreaking AI. Deepseek failed this test 91% of the time, meaning it can be manipulated to generate content it shouldn’t. This makes it a significant risk for businesses relying on AI for sensitive tasks.

Deepseek also failed 86% of the time in quick injection attack tests. This means attackers can manipulate it to leak sensitive data, bypass restrictions, or perform unauthorized actions. Wallarm researchers managed to exploit Deepseek to discover their internal systems, highlighting the severity of these vulnerabilities.

Moreover, Deepseek failed 93% of the time in malicious software generation tests. This means it can help users create harmful scripts and viruses, making it a dream come true for cybercriminals. AI models are supposed to have protection measures against offensive, discriminatory, or harmful content, but Deepseek failed these tests 68% of the time. This raises serious concerns about its reliability and ethical use.

Deepseek also struggles with hallucinations, making up information 81% of the time. If you can’t trust the information it provides, how can you rely on it for critical tasks?

Former NVS Hacker Jake Williams points out the significant difference between open-source AI and open-source code: “It is important to remember that open-source AI (such as Deepseek’s R1) means something different than open-source code. Using open-source code, we can audit the code and identify the vulnerabilities. Using open-source AI, we can’t do any of that. There are also very real concerns about the supply chain, R1 is quite easy to jailbreak and has much less guard than other commercial models.”
Jake Williams

Impact on Your Business

Deepseek is an attractive AI option, especially for small businesses or individuals who may not have the budget for more expensive models. It is low-cost, open-source, and almost as good as competitors like OpenAI’s ChatGPT and Google Gemini. However, before integrating it into your business, you need to ask yourself: is it worth the risk?

The Appsoc report suggests that Deepseek is a ticking time bomb with legal, security, and financial obligations waiting to explode. Unlike OpenAI, Microsoft, and Google, which offer legal protection, Deepseek does not support its users. This means you are on the hook if something goes wrong.

High levels of hallucination failure make Deepseek unreliable for tasks requiring accuracy, such as financial analysis, legal guidance, or medical applications. The risk of supply chain failure also raises concerns about data integrity. If companies don’t know where Deepseek’s training data comes from, should they trust it?

Consider the legal and regulatory landscape surrounding Deepseek. Some countries and governments have already banned or restricted its use. If your business relies on Deepseek today, it could be forced to abandon it tomorrow, causing disruption and wasted resources.

NowCure Security Expert Andrew Hoog also found security flaws in the Deepseek iOS app, which did not encrypt transmitted data. This opens the door for data theft. Talking to Brian Krebs on Krebsonsecurity, Hoog openly stated:

“When we see people show really simplified coding errors when you dig deeper, there are usually many more problems. There is virtually no priority for security or privacy.”
Andrew Hoog

The Chinese Connection

Reports indicate that Deepseek may contain hidden code sending user data back to China. Ivan Tsarynny, CEO of Feroot Security, warns:

“Our personal information is sent to China, there is no denial, and the Deepseek tool collects everything American users connect with it.”
Ivan Tsarynny

Security firms have found direct links between Deepseek and Chinese government servers, raising questions about data monitoring and potential data theft. Businesses using Deepseek risk becoming unwitting data repositories feeding sensitive information to foreign entities.

NordVPN Cyber Security Expert Adrianus Warmenhoven points to Deepseek’s data collection policy:

“This raises concerns about the collection of described data—starting from user-shared information to data from external sources, which are at potential risks related to storing such data in jurisdictions with various privacy and safety standards.”
Adrianus Warmenhoven

Deepseek also adheres to Chinese rules and laws related to content. Investigators have discovered that Deepseek censors politically sensitive topics and aligns its answers with Chinese state narratives. This raises concerns about its potential as a propaganda tool.

Conclusion

AI should serve as a tool, not a liability. While Deepseek offers advanced capabilities at a fraction of the cost compared to its competitors, the real costs—such as data exposure, compliance risks, and geopolitical concerns—could be much higher.

The most disturbing part is that Deepseek’s deficiencies are not potential scenarios. These are fundamental problems that make Deepseek a very real liability for companies that choose to use it. We are talking about an AI model that can be manipulated, exploited, and potentially used for cybercrime, exposing companies to serious legal and financial consequences.

So, before integrating Deepseek into your business operations, assess the risks against the rewards. This is not just about choosing an AI model; it’s about deciding whether to risk the security, reputation, and future of your company. While Deepseek may save money compared to other AI models, its true cost could be much higher. After all, there’s no such thing as a free lunch.

DeepSeek: the Ethical and Security Concerns Surrounding the AI Revolution

What are the Key Concerns Regarding DeepSeek’s Security and Ethics?

DeepSeek, despite its rapid popularity in the tech industry, has faced important scrutiny from a recent report by Appsoc. This report raises alarming concerns about the model’s security and ethical standards, suggesting it may not be as reliable or secure as claimed. These parallels to george Orwell’s “1984” highlight the potential risks for both businesses and casual users.

Why Has DeepSeek Become So Popular?

originally emerging as an underdog in the AI industry, DeepSeek gained rapid traction by offering an open-source model that is highly modifiable and free to use. This drastically undercuts competitors like OpenAI’s ChatGPT, which charges ample monthly fees.

DeepSeek, backed by substantial Chinese venture capital, disrupted the AI market, leaving significant economic impacts, particularly on the U.S. technology stock market.

What Hidden Dangers Does DeepSeek Pose?

As the Appsoc report points out, DeepSeek can be likened to an intern who, despite appearing competent, commits costly errors. It can be manipulated to leak sensitive data, generate malicious software, and bypass ethical content measures, thus posing substantial risks to businesses.

Why is DeepSeek Considered “Deeply Mistaken”?

The technical aspects of the Appsoc report indicate that DeepSeek failed numerous security tests:

• 91% of jailbreak tests
• 86% of speedy injection attack tests
• 93% of malicious software generation tests
• 68% of content protection measures

DeepSeek’s propensity for misinformation (81% hallucination failure) further questions its reliability for critical tasks.

How Does the Nature of Open-Source AI Differ from Open-Source Code?

Jake Williams, former NVS Hacker, emphasizes the differences between open-source AI and traditional open-source code:

“It is important to remember that open-source AI (such as Deepseek’s R1) means something diffrent than open-source code. Using open-source code, we can audit the code and identify the vulnerabilities. Using open-source AI,we can’t do any of that. There are also very real concerns about the supply chain, R1 is quite easy to jailbreak and has much less guard than other commercial models.”

Jake Williams

What Impact Could DeepSeek have on Your Business?

While DeepSeek offers affordable AI solutions,its security flaws could expose businesses to:

• Legal and compliance issues
• Data integrity concerns
• Potential requirement to abandon the tool suddenly due to regulatory actions

Moreover,the lack of legal protection means businesses bear the full brunt of any issues arising from DeepSeek’s vulnerabilities.

What Security Flaws Have Been Found in deepseek’s iOS App?

Security expert Andrew Hoog identified critical flaws in the DeepSeek iOS app, notably its lack of data encryption.This vulnerability can facilitate data theft, as highlighted by Hoog in a discussion with krebsonsecurity:

“When we see people show really simplified coding errors when you dig deeper, there are usually many more problems. there is virtually no priority for security or privacy.”

Andrew Hoog

Is There a Connection Between DeepSeek and the Chinese Government?

Investigations indicate DeepSeek may contain code that sends user data to Chinese servers. Security experts like Ivan Tsarynny and Adrianus Warmenhoven express concerns over user data potentially being fed to foreign entities and highlight ethical concerns regarding content censorship aligned with Chinese narratives.

What Should Businesses consider Before Adopting DeepSeek?

While the allure of low-cost AI is strong, businesses must weigh potential risks against benefits. The fundamental issues with DeepSeek transform it from a convenient tool into a liability, potentially exposing organizations to severe consequences.

integrating DeepSeek into operations requires careful risk assessment. Understanding the gravity of data exposure, compliance challenges, and geopolitical implications is crucial before making a decision.