Experts Forecast AI, SaaS and Identity Risks to Shape Cybersecurity in 2026
Summary
Darktrace experts predict that AI agents, prompt injection, AI-enabled cybercrime, human-targeted scams, and attacks on SaaS providers will intensify cybersecurity risks in 2026.
Key points
- AI agents may expose data or take unintended actions when influenced by prompts or indirect prompt injection; the article recommends treating agents as identities and restricting and monitoring their behavior.
- Experts expect indirect prompt injection against AI chatbots or agents to cause unauthorized data exposure or actions outside intended scope.
- Deepfakes and fast-moving scams are making human judgment less reliable as a last line of defense; proposed safeguards include provenance checks, cryptographic signatures, and dual-channel verification.
- The experts predict AI and automation will increase attackers’ capacity, potentially bringing more organizations into reach.
- SaaS providers may face increased supply-chain targeting, with attackers using valid credentials, APIs, or misconfigurations to bypass traditional defenses.
- Cybercrime playbooks for misusing or jailbreaking AI models may become more commercialized, scalable, and reusable.
Article Details
- Topic
- Predicted cybersecurity trends for 2026, including agentic AI risks, prompt injection, AI-assisted cybercrime, and SaaS supply-chain targeting.
People
Collin ChapleauSenior Director of Security & AI Strategy; predicts major indirect prompt-injection incidents involving companies using AI.Margaret CunninghamVP of Security & AI Strategy; discusses human vulnerability to machine-speed cyber operations and the need for defensive safeguards.Max HeinemeyerGlobal Field CISO; discusses how AI and automation may increase attacker capacity and affect smaller organizations.Nathaniel JonesVP of Security & AI Strategy; predicts increased targeting of commercial SaaS providers.Nicole CarignanSVP of Security & AI Strategy; discusses agentic AI as an emerging insider risk.Toby LewisGlobal Head of Threat Analysis; discusses the commercialization of AI-assisted cybercrime.