An AI security research and products company, Glemad, has launched a new artificial intelligence model designed to autonomously detect and respond to cyber threats, as organisations seek faster and more efficient ways to secure digital infrastructure.
The company disclosed this in a press statement on Sunday, introducing the system known as Autonomous Defence Transformers.
The statement read, “Artificial intelligence has transformed how people write, search, and code. A new system called ADT is now attempting to do the same for how organisations defend themselves online, and it works nothing like the AI tools most people are familiar with.”
According to Glemad, the model was developed specifically for cyber defence and differs from general-purpose AI systems by focusing solely on protecting infrastructure in real time. It is designed to monitor systems, analyse activities, and respond to threats as they occur.
Founder and Chief Executive Officer, David Idris, said the system was built to act at the speed required by modern digital environments. He stated:
“Defence must operate at the same speed as the systems it protects. We built ADT to act, not to inform.”
The firm explained that the current version, ADT-4, is built around what it describes as persistent reasoning, allowing the system to accumulate context over time and develop an evolving understanding of normal activity across an organisation’s infrastructure.
When deviations from this baseline occur, the system classifies the threat, evaluates the appropriate response within predefined policy limits, and automatically executes containment actions. “Access restrictions, process terminations, and attack path isolations happen in milliseconds, without waiting for human approval,” the statement noted.
Glemad also said all decisions taken by the system are logged with traceable reasoning, enabling operators to audit why specific actions were taken.
Providing performance details, the company reported that ADT processes 45,000 security events per second, has a mean detection time of 0.8 minutes, resolves 95 per cent of confirmed threats autonomously, and achieves a threat classification accuracy of 97 per cent. The technology has already been deployed to protect more than 680,000 digital assets.
The company said the model is available through its enterprise platform, PulseADT, which integrates directly into existing infrastructure. An application programming interface has also been released to enable organisations and developers to embed the system’s capabilities into their own platforms.
Glemad noted that the research underpinning the model is publicly available across preprint platforms and academic indexes, describing ADT as a new model class built specifically for adversarial environments. The statement added that as digital transformation accelerates, organisations are facing expanding attack surfaces, and the model is designed to respond at machine speed.
The PUNCH previously reported in December 2025 that Nigerian organisations faced the highest volume of weekly cyberattacks in Africa, according to the African Perspectives on Cyber Security Report 2025 by Check Point Software Technologies Ltd., a global cybersecurity leader.
The report revealed that Nigerian firms experience an average of 4,200 attacks per week, significantly higher than the continental average of 3,153 and 60 per cent above the global average of 1,963 attacks per organisation. The findings highlight a sharp rise in attacks across Africa, driven largely by AI-enabled threats.
Country Manager for West Africa at Check Point, Kingsley Oseghale, said attackers are increasingly using AI to automate phishing, impersonation, and cloud exploitation. The report urged African businesses and governments to adopt prevention-first security strategies, including continuous risk assessment, regulatory readiness, and public-private collaboration.
