Artificial intelligence is becoming one of humanity’s most powerful tools, promising breakthroughs in medicine, science, education and productivity. But developments during September have intensified a very different question: what happens when increasingly autonomous AI systems are used for cyberattacks, espionage, fraud and weapons development?
From assistant to operator
A new threat-intelligence report from Anthropic provides one of the clearest indications of how quickly the technology is changing.
The company says it disrupted operations in which AI was being used not simply to answer questions or generate code, but to help orchestrate cyber operations. In some cases, multi-agent systems carried out reconnaissance, exploitation and data processing while humans selected targets and reviewed the results.
Anthropic says suspected state-sponsored groups, financially motivated criminals and other actors have attempted to exploit its Claude models.
One Russian-linked operation used AI to accelerate cyber-espionage activity targeting Ukrainian and European government organisations, diplomatic and defence targets and people connected to US foreign policy.
The barrier to sophisticated attacks is falling
Perhaps the most important development is accessibility.
Anthropic argues that AI is reducing the gap between highly sophisticated cyber organisations and much smaller groups or individuals. Tasks that once required teams of experienced specialists can increasingly be accelerated or partially automated.
The Bank for International Settlements has identified the same dual effect: frontier AI can strengthen cyber defence while simultaneously increasing the speed, scale and complexity of attacks.
But AI is also becoming an extraordinary tool
The other side of the equation is equally significant.
Google says AI is already being developed to accelerate scientific research, improve disease detection, predict floods and wildfires, overcome language barriers and support education. The potential economic and humanitarian benefits could therefore be enormous.
Anthropic CEO Dario Amodei has similarly argued that advanced AI could dramatically accelerate medical discoveries and economic growth, while simultaneously calling for greater caution over the pace at which frontier systems are developed.
Tool or threat may be the wrong question
AI itself is increasingly becoming both.
The same capabilities that allow an AI system to search millions of lines of software for vulnerabilities can help defenders repair them — or help attackers identify weaknesses. Systems capable of accelerating pharmaceutical research can also potentially provide knowledge that requires safeguards.
The critical question is therefore shifting from whether artificial intelligence is beneficial or dangerous to whether governments, developers and society can maintain effective human control as its capabilities increase.
That race may become one of the defining technological challenges of this decade.
Newshub Editorial in Europe – 20 September 2026
Ask NF GPT
If you have an account with ChatGPT you get deeper explanations,
background and context related to what you are reading.
Recent Comments