AI and cyber risk: key takeaways from Microsoft’s 2026 Digital Defense Report
Please confirm you are human to start the audio.
Key contact
Microsoft has published its 2026 Digital Defense Report, which looks at defending organisations in an interconnected world. For general counsel, compliance teams and boards, the message is clear: cyber risk is a continuous test of business resilience, not just a technical problem for IT departments. An executive summary is also available to read here.
The report’s AI section opens with a stark message: AI is “changing the physics of cybersecurity… reducing the time, expertise and cost required to discover and exploit weaknesses”. AI is now both a tool and a target for attackers. The figures make the point:
- 88% of enterprises are already experimenting with AI agents.
- The median time from a vulnerability being discovered to being weaponised has fallen to well below 24 hours, with a record of around 72,000 Common Vulnerabilities and Exposures estimated to be tracked in 2026.
- Malicious browser extensions posing as AI assistants have reached nearly 900,000 installs and affected more than 20,000 enterprise tenants.
The report also tracks the move towards fully autonomous attacks. July 2026 saw the first documented automated ransomware extortion attack, and the report expects agentic systems soon to assess and exploit exposed enterprise assets continuously. For now, most observed campaigns still retain meaningful human direction, but the report warns that this is unlikely to last for very long.
For legal and governance teams, the most significant findings concern AI agents deployed inside organisations. The report identifies three risks: agents proliferating faster than organisations can track them, agents accumulating access that is rarely revoked, and no human being accountable when an agent acts autonomously.
A number of the AI recommendations fall squarely within the remit of legal and risk functions and provide useful insights for in-house legal and risk teams.
These findings have direct consequences for AI governance frameworks, accountability for autonomous systems, data protection, and the AI terms in supplier and technology contracts. The report recommends that organisations without a suitable forum form a dedicated agent governance committee, including compliance specialists and agent sponsors. As the report puts it, “the choices made today will determine whether AI becomes a force multiplier for cyber defense or a catalyst for systemic cyber risk”.