Ought AI improves the assurance and survivability of fully autonomous systems.

Think of the fuse in a plug or outlet. A fuse protects against accidents and we do the same job for autonomy. Our product monitors the AI that controls autonomous systems to provide vulnerability warnings, so we don’t need to intervene most of the time.

But in safety critical scenarios, Ought AI can automatically control power at machine speed to disable compromised autonomous systems and render them safe – just like a fuse. Our solution gives manufacturers hardened assurance and regulatory compliance against threat intervention and technical failures that lead to unintended consequences – enabling full autonomy to be deployed safely.


Increasingly agentic and general AI brings progress with new challenges like harmful goals or deceiving supervisors. Adversaries are keen to attack these vulnerabilities and subvert autonomous systems using novel approaches. Assurance of AI and autonomous systems is a shared global challenge because meaningful human control over autonomy must be maintained.

Until a generally robust fail-safe human control is available, more convenient, cheaper and safer mobility cannot scale safely. Beyond civilian applications, defence forces face being overwhelmed by novel threats of materially different threat volume and speed where in-loop methods risk being ineffective. Approaches for meaningful control must keep pace with innovation.


Defence spending with UK industry by the Ministry of Defence has topped £25bn for the first time and the US DoD budget stands at $841bn for 2024. Global defence spending increased by 9% to a record $2.2tn during 2023. By advancing assurance innovation, Hybrid Vision has secured granted patents for the OughtAI computer-implemented invention in the UK (GB2588697), US (11,733,697) and Australia (2021215370) in support of AUKUS.

Ought AI technology is exploitable by selected partners and our company mission prioritizes continued UK and allied security and prosperity. Make contact to learn more.