A cybersecurity briefing on third-party compromise, software dependencies, and the accelerating role of Frontier AI.
Supply chain attacks have moved from specialist cyber-risk discussions into the mainstream of business resilience. In 2025, adversaries increasingly stopped trying to break through an organisation’s front door and instead targeted the trusted technology relationships sitting around it: software vendors, open-source packages, SaaS platforms, managed service providers, identity providers and CI/CD pipelines. The result is a powerful shift from direct compromise to indirect compromise. One weak supplier, poisoned dependency or abused update channel can give attackers access to many downstream victims before defenders realise that the original point of entry was outside their own perimeter.
Quick Takeaways
- Supply chain compromise increased because enterprise environments now depend heavily on interconnected vendors, APIs, cloud platforms and open-source components.
- Software registries became a high-scale attack surface, with npm remaining a dominant vector for malicious package activity.
- By 2026, supply chain risk is no longer only an IT issue; it is a resilience, procurement, identity and governance issue.
- Frontier AI is accelerating parts of the attack lifecycle, but it is a force multiplier rather than the main cause of the trend.
Why did supply chain attacks increase in 2025?
The main reason is that the modern enterprise attack surface became more connected. Attackers gained more routes through vendors, APIs, support portals, cloud services, software updates and developer tooling. IBM’s 2026 threat analysis says major supply chain and third-party incidents quadrupled over five years, showing how attractive trusted integrations have become compared with hardened corporate front doors.
Open-source ecosystems also became a high-scale delivery mechanism. ReversingLabs reported a 73% rise in malicious open-source packages in 2025, with activity concentrated heavily in npm. At the same time, credential theft, weak MFA, leaked developer secrets and compromised maintainers made it easier for adversaries to tamper with build systems or publish malicious dependencies.
Where does the risk stand in 2026?
The 2026 picture is not a temporary spike; it is a structural change in the attack surface. CrowdStrike’s 2026 Threat Hunting Report, based on frontline observations from July 2025 to June 2026, says adversaries are exploiting trusted access paths across identity, cloud, SaaS, endpoint and perimeter infrastructure. This creates cross-domain intrusions that are difficult for fragmented security teams to connect quickly. CrowdStrike also reports that npm, the default package manager for Node.js and the world’s largest software registry, remained the primary vector for software registry abuse, accounting for 87% of malicious software registry threats in the first half of 2026. In one example, ALTERED SPIDER compromised more than 300 software dependencies in a single day before harvesting credentials and pivoting into cloud environments.
What role does Frontier AI play?
Frontier AI is best understood as a force multiplier, not the root cause. CrowdStrike reports that AI-enabled adversary activity surged 89% in 2025 and that, from January to June 2026, 88% of observed exploitation of vulnerabilities with a public proof of concept occurred within 48 hours of release.
For supply chain attacks, that matters because AI can compress the time between discovery, weaponization and downstream compromise. It can help attackers with reconnaissance, phishing, code analysis, exploit development and mapping attack paths through code repositories, CI/CD pipelines, IDE extensions and cloud identities. The risk is less about fully autonomous attacks today and more about speed, scale and precision.
Wrap-up: What should organisations prioritise?
- Require software bills of materials and signed builds from critical suppliers.
- Harden CI/CD pipelines, developer workstations and package publishing workflows.
- Use hardware-backed MFA for maintainers, administrators and support accounts.
- Continuously scan for leaked secrets, risky dependencies and abnormal package behaviour.
- Include supplier compromise and SaaS outage scenarios in incident response exercises.
