I know it is summer and we ought to try and look on the brighter side of life by focusing on more positive and hopeful topics. But the late-summer silly news season is nowhere to be seen. Instead, we have wildfires and drought plaguing Europe, accompanied by the ongoing belief that the world is yet to see the full ramifications of the US-Iran war. And not to forget the intensified and relentless Ukraine-Russia war. Add to that mix the never-ending artificial intelligence revelations: that one day the machine will take over human jobs, followed by humans betting on machines becoming autonomous or even more intelligent than those who invented them.

What is troubling is that the writing on the wall about the excesses of the “intelligent machine” are everywhere, yet it feels retrogressive to vocalize those doubts in public, let alone in an intelligent conversation calling for regulation. This is not to save the mundanity of human life but to ensure its continued relevance and to ward off annihilation. Machines ought to be at the service of a better humanity, not one that is exploited or twisted by an algorithm that cancels out humans and their existence.

A few recent incidents and revelations in the realm of AI have created cause for concern, even if they have been made to either spread fear or justify yet another round of funding worth billions if not trillions of dollars to further develop, exploit and monetize this new tech frontier. Some believe the growing concerns about AI are themselves a marketing ploy, but surely the revelations are legitimate and reflect people’s concerns.

The first was that OpenAI recently revealed that one of its own experiments had led the machine to autonomously hack a fellow AI company. Anthropic followed suit to reveal that its Claude system had apparently done the same and advised competitors to check their systems. Meta later rushed to suggest that its researchers had found evidence of similar behavior. These announcements seem to confirm our most extreme sci-fi fears: Despite their many services to humanity and importance to our future, these powerful systems can go rogue.

Despite their many services to humanity and importance to our future, these powerful systems can go rogue

Mohamed Chebaro

The second was the announcement that AI had been used to design brand new viruses that are fully functional and can replicate in the laboratory. The resulting 16 novel viruses were apparently created to infect bacteria and pose no threat to humans. This turning point in science could unlock a new era in the treatment of diseases, according to the researchers. But experts have warned that AI-designed viruses raise “urgent” safety and security concerns due to their potential malicious use.

The third concerns revelations about weapons of war achieving enhanced lethality when combined with AI. The world is yet to comprehend the effects of fully autonomous systems that, if let loose, could result in disasters, mistakes and the misidentification of targets, causing harm to civilians. Due to AI, modern warfare rests on autonomous weapons systems that are capable of limitless engagement of targets, potentially not excluding civilians of course. In this new environment, one wonders if the Geneva Conventions are still applicable as the guardrail that has for decades forced armies to respect and uphold the principles of distinction, proportionality and meaningful human control.

The question is whether this is still the case when AI is delivering an array of new tools that make weaponry more potent, from uncrewed drones to a computer controlling a swarm of advanced drones that can engage hundreds of targets without human interference.

Existential risks are the key worry here, away from marketing or the need to fine tune the technology or even regulate it. AI companies have, at least in the Western world, been quite open about the fact that their models hallucinate — maybe to say that they need more resources or to limit their accountability. But what about other programs driven by no less aggressive entities like states and nonstate actors in the total darkness, away from all transparency?

Existential risks are the key worry here, away from marketing or the need to fine tune the technology or even regulate it

Mohamed Chebaro

Recent disclosures have brought about concerns that echo the need for regulatory attention. And the White House’s call for a conversation with the tech giants about their testing procedures is a good start.

Yet more needs to be done by all actors in this field. No one dares talk about this because it could ruin the exploitative model that these tools’ development has rested on from the onset.

One of the worries is that AI companies always circumvent the need to moderate content more efficiently and help avert misinformation, disinformation and fake news, which can rip societies apart. Above all, AI must be prevented from reducing human knowledge through its immense quantity of slop, which undermines learning and critical thinking — even if it is against the commercial interests of the tech giants.

The world does not need more evidence of the dangerous, unregulated and often irresponsible expansion of AI. The case of systems going rogue or tools hacking other tools could become the name of the game, heralding a tech wild west. That is unless the world and its leaders dare to have a sober conversation about overdue regulations. Or they could agree a moratorium similar to the global protocols of nonproliferation that still guide the use of nuclear research and weapons in order to prevent AI tools falling into the wrong hands and becoming cyber weapons of mass destruction.

  • Mohamed Chebaro is a British-Lebanese journalist with more than 25 years of experience covering war, terrorism, defense, current affairs, and diplomacy.