Artificial Intelligence serving National Security: the Spanish case

In this paper, which is part of a work in progress, we want to analyse and underline the contradiction that arises in the societies of knowledge between the opacity of most AI systems and the transparency that, as a tool for good governance, is been looked for by the most progressive political, social and economic leaders. On the one hand, the risks and threats for the principles of good governance (among them, transparency) derived from the use of AI by governments in sensitive areas such as that of National Security. While AI is to make it easier and more efficient the work of Security Forces, our democracies will become weaker and less credible if the protection of our rights and principles becomes lower than it is thanks to these AI systems, with a risk for autocracies to arise. Thus, our goal should be to produce National Security AI systems that are ethic and legal in a moment of history in which transparency is considered essential for good governance. On the other hand, the use of new disruptive technologies to protect State secrets only makes it more and more difficult to monitor the policies of our governments. The fact is that transparency becomes a reality by exercising the right of access to public information. If we are not allowed to properly monitor our leaders, as a matter of fact protected by smart shields such as those AI systems, we would have taken one step back. Our aims would be carefully identifying the threats to our principles and rights, in the first place, and, secondly suggesting the ethic and legal traits of AI systems to be considered by policymakers that are currently amending State secrets regulations In the case of Spain, a lack of coherence pervades the corresponding legislation, if we take into account that the Official Secrets Law is dated in 1968, nothing to do with the last National Strategy of Security (2021). The lack of consistency between old regulations and new disruptive AI in Spain claims for quick and deep updating.