London, 27 October 2026 – New international guidance has been issued to help organizations that develop or use AI to identify and address the security threats specific to AI systems. It comes as research covering 1,287 leaders globally suggests businesses see cyber risk as the biggest AI risk to their organization, ranked well ahead of skills gaps, data transparency, accountability or bias.
The guidance, developed by ISO/IEC and published in the UK by BSI, the UK's national standards body, recognizes that AI is now becoming embedded into business operations, with BSI’s study showing that agentic AI use is already widespread, with nine in ten companies exploring it in some form. Yet while established cybersecurity controls remain relevant, these are likely to need adapting or supplementing to address new risks.
While many organizations will be using AI to enhance their security, evidence suggests cybercriminals are exploiting vulnerabilities in AI systems and attacking them, while also using AI to facilitate conventional cybercrime. Tactics can include manipulation of training data and inputs, theft or reconstruction of models, and unintended behaviour triggered through prompts or outputs. The UK’s National Cyber Security Centre has predicted that AI will lead to an increase in frequency and intensity of cyber threats.
Cybersecurity. Artificial intelligence (AI). Addressing security threats and compromises to AI systems (BS ISO/IEC 27090:2026) explains how attacks and vulnerabilities can affect AI data, models, outputs and services throughout the AI system life cycle, describes their potential consequences, and sets out approaches to detecting and mitigating the risks. It covers threats including data poisoning, prompt injection, model theft and leakage of training data or model inputs. The guidance applies to organizations of all types and sizes, including public and private companies, government bodies and not-for-profit organizations.
BSI’s study found that security was the top concern around agentic AI, with more than half of business leaders (53%) stating this. AI cybersecurity was also the most common answer when asked which AI capability would become more important in the coming year, cited by 48%. The research, which explored the cost of AI failures, also found that 75% of organizations had experienced an AI failure in the past year, with 30% of those affected saying it had led to a security breach.
With 78% of businesses expecting their AI investment to increase, and deployment moving steadily from pilots to scale, exposure to security risk is likely only to grow. BS ISO/IEC 27090 maps threats to the assets they affect and the life-cycle stages where they arise, from design and training through to deployment and operation, and combines conventional cybersecurity controls with AI-specific measures.
Laura May Bishop, Sector Lead, AI and Cybersecurity, BSI, said: “For all the opportunities posed by AI, there is no doubt it introduces serious security risks that conventional cybersecurity measures alone were not designed to address. Cyber criminals can be relied on to ruthlessly exploit AI, so business leaders desperately need to stay one step ahead. We are already seeing instances of AI failure leading to security breaches – and it’s likely the true scale of the challenge is even greater.
“The new standard, BS ISO/IEC 27090 gives organizations around the world a structured way to understand these threats across the whole life of an AI system and to decide on proportionate safeguards. Securing AI is not only a job for developers. It needs security, risk, governance and procurement teams working together.”
The standard is designed to work alongside frameworks organizations may already use. For example, it sets out how to identify AI-specific risks and safeguards that can feed into information security management under ISO/IEC 27001 and ISO/IEC 27002, and AI governance under ISO/IEC 42001, rather than creating a separate management system. It also complements ETSI EN 304 223, which sets baseline cybersecurity requirements for AI models and systems.
While BSI’s data shows progress being made, AI tools are regularly deployed by businesses without appropriate governance programmes in place, including to protect against security issues. For example, only 38% of business leaders said their organization has established an AI governance programme. The guidance treats AI security as an organizational issue rather than solely a technical one for developers, giving security, AI engineering, risk, governance, procurement and operational teams a common language for discussing and managing AI security risk. By building a better understanding of the security risks to AI systems, organizations can raise awareness across their teams, strengthen their defences and develop a stronger security culture.
BS ISO/IEC 27090 was developed through the international, consensus-based standards process by ISO/IEC JTC 1/SC 27, Information security, cybersecurity and privacy protection, and mirrored in the UK by BSI committee IST/33/4, Security controls and services, drawing on experts from industry, government, academia and the security community.
For more information on the standard, please visit: BS ISO/IEC 27090:2026 | 31 Oct 2026 | BSI Knowledge
