The race for artificial intelligence: Josh Lospinoso's alarm

Before founding his current startup, Shift5, which works with the US military, railway operators and airlines such as JetBlue, Josh Lospinoso had already created a cybersecurity startup that was purchased in 2017 by Raytheon/Forcepoint. A former Army captain and 2009 West Point graduate, Lospinoso spent more than a decade developing hacking tools for the National Security Agency and the US Cyber Command.

Threats to artificial intelligence and the protection of military operations

Recently, Lospinoso spoke to a Senate Armed Services subcommittee on the top threats to artificial intelligence (AI)-based technologies: data theft and "data poisoning." He explained the latter as a form of digital disinformation in which adversaries manipulate the data seen by AI technologies, profoundly affecting how they function. He cited the example of Microsoft's Tay chatbot, which started generating offensive content in 2016 due to data manipulation by malicious users.

The use of artificial intelligence in cybersecurity and the inclusion of AI in weapons systems

Lospinoso then underlined the importance of artificial intelligence in the field of cybersecurity, citing its use in e-mail filters and antivirus software as examples. He also spoke on the vulnerability issue of weapons systems, citing a 2018 Government Accountability Office report that revealed critical vulnerabilities exist in most new weapons systems. Lospinoso expressed concern about the possible introduction of AI into such systems, as many of them are dated and difficult to update, and once compromised, can cause enormous damage.

Concerns about the rush to AI and the accountability of the military sector

Finally, Lospinoso expressed concern about the rush to produce AI products, noting that a hasty search can lead to serious safety problems. However, he said stopping AI research, as some have suggested, would be a mistake, as it could give competitors such as China an edge. Lospinoso concluded by stressing the importance of debate on these issues by the White House and Congress, and expressed strong opposition to the idea of using AI in military decisions, such as targeting, stating that we are still a long way from such a possibility.