Australia's CISC 2026 Annual Risk Review names AI a dominant threat multiplier for critical infrastructure
Australia's Critical Infrastructure Security Centre (CISC) released the fourth Critical Infrastructure Annual Risk Review during Critical Infrastructure Security Month 2026. It covers energy, food, water, transport, communications, health, and banking and finance. The review names trusted insiders, cyber incidents, physical threats, supply chain vulnerabilities and natural hazards as the key risks of the past 12 months. It says AI has moved from an emerging risk in 2023 to a "dominant threat multiplier" in 2026. In its view, AI increases the number of threat actors and lowers the capability they need to cause harm. This is a government strategic risk assessment. It does not report a specific incident, vulnerability or exploitation event.
Detalhes técnicos
- Frontier models are rapidly finding and exploiting vulnerabilities, and the review says they are outpacing defences. Examples cited are OpenAI agents autonomously infiltrating Australian government websites in 2026 and a state-sponsored actor manipulating Anthropic's Claude Code in late 2025.
- State-sponsored actors continue to pre-position in OT systems using living-off-the-land techniques. The review cites Iran's Seedworm (targeting US critical infrastructure since 2017) and expects AI to enhance pre-positioning.
- Concentrated dependence on a few third-party providers creates systemic risk, as the CrowdStrike and Canva outages showed. Foreign ownership, control and influence (FOCI) over a dominant provider could enable persistent access or surveillance.
- Legacy systems remain highly vulnerable, and quantum computing will worsen the risk.
- Supply chains are fragile. The closure of the Strait of Hormuz had flow-on effects on products such as bitumen, sulphur, plastics, insulation material and OT.
- AI-fuelled mis- and disinformation is eroding public trust and complicating incident communication.
Remediação
- Adopt AI-assisted detection and response.
- Review pre-positioning mitigations continually, including air gapping and restricted connectivity between OT, IT and the internet.
- Map dependencies beyond linear connections, and apply the FOCI Risk Assessment Guidance to vendors that are business critical, hard to replace, highly connected or able to collect data.
- Comply with the Enhanced CIRMP Rules on third-party and legacy system risk.
- Put governance guardrails on AI use in operations.
- Invest in workforce development and succession planning.
- Prepare for quantum risk by prioritising sensitive data, assessing cryptographic dependencies and developing a transition plan.
- Follow ASD's CI Fortify guidance on isolating vital OT.
Linha do tempo
-
Australia's CISC 2026 Annual Risk Review names AI a dominant threat multiplier for critical infrastructure
ler a análise desta fonte ocultar
Australia's Critical Infrastructure Security Centre (CISC) released the fourth Critical Infrastructure Annual Risk Review during Critical Infrastructure Security Month 2026. It covers energy, food, water, transport, communications, health, and banking and finance. The review names trusted insiders, cyber incidents, physical threats, supply chain vulnerabilities and natural hazards as the key risks of the past 12 months. It says AI has moved from an emerging risk in 2023 to a "dominant threat multiplier" in 2026. In its view, AI increases the number of threat actors and lowers the capability they need to cause harm. This is a government strategic risk assessment. It does not report a specific incident, vulnerability or exploitation event.
Technical Details
- Frontier models are rapidly finding and exploiting vulnerabilities, and the review says they are outpacing defences. Examples cited are OpenAI agents autonomously infiltrating Australian government websites in 2026 and a state-sponsored actor manipulating Anthropic's Claude Code in late 2025.
- State-sponsored actors continue to pre-position in OT systems using living-off-the-land techniques. The review cites Iran's Seedworm (targeting US critical infrastructure since 2017) and expects AI to enhance pre-positioning.
- Concentrated dependence on a few third-party providers creates systemic risk, as the CrowdStrike and Canva outages showed. Foreign ownership, control and influence (FOCI) over a dominant provider could enable persistent access or surveillance.
- Legacy systems remain highly vulnerable, and quantum computing will worsen the risk.
- Supply chains are fragile. The closure of the Strait of Hormuz had flow-on effects on products such as bitumen, sulphur, plastics, insulation material and OT.
- AI-fuelled mis- and disinformation is eroding public trust and complicating incident communication.
Remediation
- Adopt AI-assisted detection and response.
- Review pre-positioning mitigations continually, including air gapping and restricted connectivity between OT, IT and the internet.
- Map dependencies beyond linear connections, and apply the FOCI Risk Assessment Guidance to vendors that are business critical, hard to replace, highly connected or able to collect data.
- Comply with the Enhanced CIRMP Rules on third-party and legacy system risk.
- Put governance guardrails on AI use in operations.
- Invest in workforce development and succession planning.
- Prepare for quantum risk by prioritising sensitive data, assessing cryptographic dependencies and developing a transition plan.
- Follow ASD's CI Fortify guidance on isolating vital OT.