Artificial intelligence is often discussed as a future security threat. In reality, its impact on warfare is already being felt.
Appearing on CNN’s Isa Soares Tonight, General Sir Richard Barrons, former Commander of the UK’s Joint Forces Command and Senior Consulting Fellow at Chatham House, discussed how AI is changing the speed, scale and accessibility of modern conflict.
The conversation followed the publication of an Anthropic report stating that the company had blocked accounts suspected of using its AI systems to support potentially harmful biological and military research. According to the report, the activity included possible attempts to develop missile and rocket systems, autonomous drone capabilities and biological threats.
For General Barrons, the central message was clear: AI-enabled warfare is not a distant possibility. It is already here.
The convergence of powerful technologies
AI does not operate in isolation. Its military significance comes from its convergence with other technologies, including:
- Vast stores of digital data
- Robotics and autonomous systems
- Satellite and mobile sensors
- Bioscience
- Cyber capabilities
- Open-source intelligence
- Social media platforms
Together, these technologies are transforming how military forces collect intelligence, identify targets, make decisions and conduct operations.
General Barrons noted that advanced armed forces already rely on AI to process information and coordinate activity at a scale that would be impossible through human analysis alone. As these capabilities become more widely available, they will increasingly influence how both states and non-state groups prepare for and engage in conflict.
Advanced capabilities are becoming more accessible
One of the most significant risks is the growing accessibility of AI.
Actors no longer need access to the most advanced frontier models to gain a military advantage. A less capable system may still be effective enough to assist with weapons design, drone operations, surveillance, cyberattacks or information manipulation.
This lowers the technical barrier for smaller states, militant organisations and other non-state actors. Capabilities that once required extensive scientific expertise, specialist infrastructure and significant funding may now become easier and faster to develop.
The Anthropic report illustrates this concern. Among the cases discussed during the interview were suspected attempts by actors in Yemen to use AI in support of missile and rocket development, as well as an alleged Russian effort involving a drone swarm capable of selecting and attacking targets without direct human involvement.
Although these cases were identified and disrupted, they may represent only a small portion of the activity taking place across the wider AI ecosystem.
Understanding the “digital kill web”
General Barrons described the emerging battlefield as a “digital kill web”: a connected system that brings together sensors, data, AI-assisted analysis and weapons.
A sensor within this network might be a satellite, ship, drone, mobile phone or open-source database. Information gathered from these sources can be combined in a cloud-based environment and analysed using AI.
The system can then:
- Detect activity in the operating environment.
- Combine information from multiple sources.
- Identify and prioritise potential targets.
- Recommend or allocate an appropriate response.
- Direct a weapon or other capability towards the target.
That response may involve a conventional weapon such as a missile or bomb, but it could also take the form of a cyberattack, an electronic-warfare action or a coordinated social media campaign.
The defining characteristic is speed. Processes that once took hours—or even days—can now be completed in minutes or seconds.
The shrinking window for human judgement
Faster decision-making can provide a significant operational advantage. It also creates a profound challenge: how can meaningful human control be maintained when the time available to assess an AI-generated recommendation has been reduced to seconds?
A person may technically remain “in the loop,” but that does not necessarily mean they have enough time, context or confidence to question the system’s output.
This raises difficult questions:
- Who is accountable when an AI-supported decision causes unintended harm?
- How should a system respond when available data is incomplete or deceptive?
- Can human oversight remain effective at machine speed?
- What safeguards are required before autonomous systems can select or engage targets?
- How can escalation be controlled when several adversaries are using automated systems simultaneously?
These are not solely technical questions. They are matters of military doctrine, law, ethics and international security.
Governance must keep pace
AI cannot simply be removed from modern warfare. The strategic benefits are too significant, and the underlying technologies are already too widely distributed.
The priority must therefore be to establish credible safeguards around its development and use. This includes stronger testing and monitoring by AI companies, clear military rules of engagement, meaningful human oversight and international dialogue on autonomous weapons.
Governments, armed forces and technology companies must also prepare for adversaries that will not follow the same standards. Effective governance must reduce the risk of misuse without depending on universal compliance.
General Barrons’ warning is ultimately about more than new weapons. AI is changing the architecture of conflict itself: who can participate, how quickly decisions are made and how easily events can escalate.
The future of warfare is not waiting to arrive. It is already taking shape—and the decisions made now will determine how dangerous it becomes.