Hacking with AI: Criminals adopt 'point and shoot' tools against banks and governments
Combest frameworks powered by AI, the so-called 'harnesses', have been linked to real attacks against companies, government agencies and banks around the world since August...

Key takeaways
- Frameworks de pentest turbinados por IA, chamados harnesses, são usados em ataques reais desde agosto.
- Os ataques 'aponte e dispare' com IA atingiram empresas, agências governamentais e bancos pelo mundo.
- O hacking com IA não explora falhas novas, mas automatiza e escala ataques contra alvos definidos.
- Empresas e desenvolvedores no Brasil precisam se adaptar à ameaça de ataques automatizados por IA.
Criminals are using invasion testing frameworks powered by artificial intelligence — known as "harnesses" — to attack companies, government agencies, and banks in different countries since August. The tool does not invent new forms of invasion: it takes known methods and turns them into an almost automatic process, in the "point and shoot" style. The report is from Wired.
What are AI harpesses in AI hacking
The base is ancient and legitimate: Vulnerability scanners, public exploits and recognition scripts that offensive security teams use in authorized intrusion tests. The novelty is the layer of artificial intelligence above this arsenal. Instead of an expert deciding manually each step — mapping the target, choosing the loophole, adjusting the exploration code, bypassing defenses —, the harness coordinates these actions with a language model in command.
The practical result is that technical knowledge is no longer the bottleneck. What once required an experienced professional starts to run as a standardized, repeatable and parallelizable flow.
Why hacking with AI 'point and shoot' changes the game
Because the problem was never just the existence of vulnerability—it was who could exploit it. Known vulnerabilities have been circulating for years on public bases and remain uncorrected in thousands of systems. Harness does not increase the catalog of loopholes; it multiplies the number of people able to exploit them and the speed with which it happens.
It is the same risk vector that motivated the alert of organs such as the NSA on AI finding faults too fast for the human rhythm of correction — a mismatch that already concerns the financial sector, as shown in IA finds vulnerabilities too fast and overloads corrections. The difference here is the criminal use of that same ability.
Who was hit by hacking attacks with AI and since
According to WIRED, cases linked to these tools began to appear in August and hit private companies, government agencies and financial institutions in different regions of the world. The report does not present a public list of victims, nor details how many incidents were directly attributed to the harnesses. The central point of the report is the pattern: varied targets, in different countries, with tools that reduce the technical requirement of the attacker.
Hacking with AI: what changes to companies and devs in Brazil
You don't have to be finger-picked to get on the radar. When the cost of attacking falls, the volume of attempts rises — and the target is no longer just the big corporation. Medium-sized companies, software providers, exposed APIs and forgotten test environments enter the account.
On the side of the builder, this reinforces three priorities: asset inventory (know what is exposed), rhythm correction compatible with offensive automation and continuous intrusion tests instead of annuals. On the side of those who hire, the message is that offensive security with AI is no longer differential — it is baseline.
What is not yet known about hacking with AI
There are missing data that would change the reading of the case: number of affected organizations, specific countries, affected sectors, assignment to criminal groups and information about how these tools are distributed or sold. There is also no evidence that the harness completely dispenses the human operator — it reduces the barrier, does not eliminate it. Until these points appear, what is safely known is essential: automation with AI leveled access to techniques that previously required years of experience.
Frequently asked questions about hacking with AI
What's an AI harness used in hacking?
It is a framework that uses artificial intelligence to orchestrate traditional invasion testing tools, automating steps such as recognition, choice of breach and execution of the attack. The structure is derived from legitimate combest software.
Did criminals create new vulnerabilities with these tools?
Nope. According to WIRED, harpesses do not create unprecedented forms of invasion — they make existing methods easier to execute, accessible to criminals with little technical sophistication and scalable in volume.
Does my company have to worry even if it's small?
Yeah. When the technical barrier falls, the number of attempts grows and opportunistic targets become worthwhile. Maintaining updated asset inventory, correcting known faults with agility and testing exposed environments are the most effective basic defenses in this scenario.
Source: www.wired.com

Written by
Manu RamalhoSou Manu Ramalho, publicitária com 16 anos de estrada conectando marcas e pessoas. Como fundadora da EME Marketing Digital, sempre busquei o marketing estratégico para gerar conexões autênticas. Aqui, mergulho na fronteira da inteligência artificial como analista de tendências. Meu foco é traduzir a complexidade de NLP, novos modelos de linguagem e papers acadêmicos para o mundo real, sempre com um olhar atento à regulamentação, ética e aos impactos sociais que essa tecnologia imprime na nossa sociedade.


