Artificial intelligence is edging closer to what experts are calling a civilisational inflection point — one where autonomous systems could act, adapt, and potentially escalate conflict faster than human decision-makers can respond. Two recent developments, taken together, have reignited a debate that once seemed confined to science fiction.

Key Developments Driving the Alarm

US-based AI company Anthropic reportedly warned that AI could begin building itself and create its own successor — a process known as full recursive self-improvement. Recent reports caution that this capability 'might increase the risks of humans losing control over AI systems.' Separately, Anthropic's AI model Claude was reportedly used by the American Army in strikes on Iran, according to media reports including Britain's The Guardian in March. Experts cited in those reports warned that AI-enabled attacks could occur faster than 'the speed of thought,' raising fears that human oversight could be bypassed entirely.

Around the same period, a video from China went viral showing a humanoid dancing robot striking a young boy across the face — an incident widely described as alarming, though isolated. Critics noted it illustrated the unpredictability of physical AI systems operating in public spaces.

The Disinformation Threat

Beyond weapons, AI poses a parallel and arguably more immediate risk through information warfare. Jatin Gandhi, senior journalist and Trust and Safety specialist, told reporters that AI agents can be deployed by bad actors 'to do that faster and create a bigger negative impact.' He warned that making AI systems autonomous effectively 'scales up disinformation' in ways that will be 'harder to fight.'

A study from the University of South Carolina, published recently, reportedly demonstrated how AI agents can coordinate disinformation campaigns without human involvement — and, critically, without being detected. Gandhi noted this 'has huge implications for democracy, trust and safety of individuals as well as society at large.' He added, with grim humour, that unlike in The Terminator, 'there won't even be an Arnold Schwarzenegger.'

Lethal Autonomous Weapons: What the UN Said

A United Nations report flagged the growing development of Lethal Autonomous Weapons Systems (LAWS) — so-called 'killer robots' — that leverage AI to identify, select, and eliminate human targets without direct human intervention. The report also raised concerns about AI-powered robotic drone swarms as a particularly dangerous frontier.

In a discussion convened by the Harvard Medical School, researchers warned that autonomous systems could lower the threshold for armed conflict, enable rapid escalation without human oversight, and create accountability gaps when machines make lethal decisions. Participants called for international regulation, transparency in AI development, and robust safeguards — framing AI in warfare as 'not just a technological challenge but a moral and humanitarian one.'

Global Regulation Efforts So Far

Governments and international bodies have begun responding, though critics argue the pace is insufficient. In November 2023, India was among 29 countries that attended the AI Safety Summit in the United Kingdom, signing the Bletchley Declaration on AI safety. Subsequently, the UN General Assembly adopted a resolution calling for 'safe, secure and trustworthy' AI systems.

Film director James Cameron, whose Terminator franchise first dramatised the risk of AI-driven apocalypse over four decades ago, has himself warned in media interviews about the real-world danger of AI deployed in weapons systems — including, he noted, at the level of nuclear arsenals.

What Experts Say Must Happen Next

Specialists urge a three-pronged response: vigilant monitoring of AI capability advances, robust regulatory frameworks where necessary, and — most critically — preemptive intervention to constrain any system that begins exhibiting superintelligent, self-directed behaviour before it can pose a systemic threat. The window for such intervention, several researchers argue, may be narrower than policymakers currently assume.