The Fear Economy:
Who Profits When Mauritania Panics Over AI?

By Sidi SOUEINA, PhD

Nouakchott May 1st 2025

Several days ago, perhaps last week by the time I released this article, I came across a post on facebook, which in my view slightly misrepresented another post made by an ex Mauritanian minister of digital transformed on linkedin. The post was regarding new Artificial Intelligence technology developed by Anthropic. A very powerful AI company which spawned thus far the best competitor of ChatGPT, namely Claude, Claude Code and other tools. 

This time Anthropic produced what could be described as AI-driven automated vulnerability discovery (ADAVD onwards) with “imminent” danger on technology infrastructure across the spectrum. It is perhaps noteworthy to mention that as of now, these AI test tools have only been given to a squad of powerful companies for closed testing.

In order to give my opinion on AP panic though and why we should not listen to the nay sayers and fearmongers about Technology, allow me to first educate a little bit on the issue of ADAVD and the great story of Anthropic corporation, and why some consider it to be at the frontier of Cyber Security..

In the story of what some might call the AI bubble, Anthropic corporations deserve  our respect for walking away from a massive paycheck with the American Departement of War, aka The Pentagon, when it said to have refused to license its AI technology to the latter. Something that would have allowed AI controlled weaponry to make kill decisions in wars, independently  of humans! 

Based on the current evidence, this only happened in science fiction movies. As far as I can remember very few companies refused a multy hundreds of millions Dollars contracts with the government based on ethical reasons. So kudos to Anthropic for daring to make such a courageous decision at a time where most Tech companies are currying favor with the Trump government day in and day out.

Now, the technology that gives Anthropic advantages in producing near perfect AI in comparison with say OpenAI and Gemini is still unclear. However, one thing is. Anthropic produces AI tools that seem to be years ahead of when it comes to  computer programming code production compared to the giants like Google, Meta and openAI. The latest of which is Claude LLM and Claude code and a few other tools which it seems enticed the ex minister to issue a daring warning about cybersecurity in Mauritania. The tools are said to have extraordinary capabilities of finding zero-days -- the weaknesses in a network. Something that is throwing experts in cyber security circles off balance. 

This raises a set of urgent questions for policymakers in Mauritania: how to interpret these tools, how to assess real risk, and how to avoid reactionary responses?

Answering these questions requires a little bit of time and an effort to simplify the cybersecurity and artificial intelligence terminology to the masses. So let me give it a go before closing this short article with my take on the whole issue.

What does it mean to find zero-days?

Anthropic produced an ADAVD which is designed to find zero-days  in a network. You see, when software programmers and technicians write code, they do so while inadvertently missing important security and logic flaws or leaving some access doors open. Some of these flows are left undiscovered or unintended for years and sometimes even decades. 

Like the Y2K bug, which is a logical error, not a security one.

In the Y2K case, only reserved two digits in the computer memory instead of 4, famously causing serious issues, some even said that nuclear ICBMs (intercontinental ballistic missiles) will launch uncontrollably when computers would fail to increase the date beyond the year 99. This happened during the 1999 transition to the twenty-first century. As turned out, none of that happened and the code was fixed one case a time.

Zero days are important because they can lie dormant until hackers, the technicians or the general public discover them. If true, this is huge, for it can be  used by hackers to access and potentially disrupt important IT systems including government, banks, especially central banks and other important public and private institutions.

Although some have expressed scepticism over its capabilities, Anthropic is said to have refused to release the software and have been given permission to test it by a few stakeholders for now. 

So here is my take.

Since its birth, all new high tech renaissance has been accompanied by risks. People have talked about Windows OS having black hole linked to the US Government! PC and Mobile phone users are told to tape the cameras with a black tape! 

Many other such dire warnings we hear all the time. We are told that even if we switch off our phones and their screens turn black that the phones are listening! We are told that Meta and tiktok are listening to us. All  the time. 

While some of these warnings might be true, it would be foolish to just put technology aside and split from the world. That would be foolish especially when it comes to the government. Economies will crumble and lives will be lost. 

If ADAVD tools can truly dismantle sensitive networks, using its zero-days amongst other detection means, it will trigger a global security crisis where no nation is an island. We are a vital node in an interconnected global economy. While we shouldn't wait idly for ‘digital vaccines‘ as in the COVID case since cyberattacks strike in milliseconds, not months—we must recognize that the international community has a vested interest in Mauritania’s stability. Our goal should not be isolated fear, but an honest, local defense that plugs into global security patches like 'Project Glasswing'—the coalition currently testing these models.

So we need to have our own solutions and especially be sure it benefits the people and not for a select few elite consultants. 

Now back to the warning.

In my view, the Mauritania government should NOT fear ADAVD tools or any other new high tech out there. We should not fear the salesmen of AI panic who want to convert government ignorance into a profitable business model. Fearmongering happens all the time, be it a media company trying to increase subscriptions or raise its advertisement fees, a technology company getting new clients, or consultants perhaps trying to get new contracts. It makes sense for all of the above to issue these dire warnings in order to cash on those fears. 

Nonetheless, I would like to remind the government that these fears are oftentimes exaggerated and baseless and surmountable. 

Closing argument

What concerns me most is the apparent overlap between ministerial responsibilities and private business interests, where companies linked to current or former ministers are reported to secure contracts from the same ministries. This pattern, often described as a revolving-door system, raises serious concerns about transparency and the integrity of public procurement. We know who these people are, we know their companies. 

God Bless Mauritania.

Sidi Soueina, PhD
Professor of Computer Science
Institute Supérieure du Numérique
President and Founder of MiFiTT Institute