INSIGHT
How to Defend Against AI-Powered Cyberattacks With Stronger Identity and Data Governance
Linsey Gallo, Jon Parrish, Josh Schmidt • November 19, 2026
Services: Cybersecurity Risk Assessments, Cybersecurity Services
A recent Microsoft Digital Defense Report describes a threat landscape in which AI has “compressed the security cycle from days to minutes.” Attackers are using the technology to find vulnerabilities faster, build attacks with less specialized skill, and automate steps that once took a person at a keyboard. Microsoft’s advice for defenders leans on the basics: tighter control over who and what has access to your systems, and better governance of the data your AI tools can reach.
If you lead security, IT, or finance at your organization, those two priorities deserve a place on your agenda before year end.
AI Is Shrinking Your Window to Respond
Vulnerability discovery is one of the areas where AI has the most visible effect. Microsoft reports that the time between a vulnerability’s discovery in the wild and its active weaponization can be well below 24 hours, while the number of publicly disclosed vulnerabilities is projected to reach a record 72,000 in 2026. Discovery and exploitation are outpacing fixes, and Microsoft tells organizations to prepare for “a multi-year period where the number of known but unpatched vulnerabilities spikes.” Well-funded groups, including those working for national governments, may also stockpile zero-day vulnerabilities found with AI tools.
For your team, that changes the patching conversation. Few organizations will fix everything quickly, so the priority becomes knowing which exposed systems matter most and limiting how far an intruder can travel if one of them is compromised.
Attackers Are Automating More of the Attack
Microsoft pointed to the first documented ransomware operation powered entirely by agentic AI, discovered in July, and said it has observed AI-orchestrated intrusions that share elements with that activity. Volumes remain low, according to the report, but the activity is not confined to a single sector or region.
Familiar tactics are getting a boost as well. Phishing was the way in for 23 percent of the intrusions Microsoft’s incident responders investigated between July 2025 and June 2026, up from 7 percent a year earlier, and exploits against public-facing applications rose from 15 percent to 24 percent. AI lets attackers personalize every phishing message and helps them get past language and skill barriers, so the awkward phrasing and obvious mistakes your colleagues have been trained to spot are becoming less reliable.
Identity Is Your Primary Control Plane
Microsoft states it plainly: “Identity is the primary control plane for defense.” Attackers regularly take advantage of unnecessary privileges, persistent access, and weak identity governance. In more than half of intrusions that began with valid accounts, they harvested additional credentials once inside, which shows how quickly one compromised login can become many.
The safeguards Microsoft recommends will sound familiar: phishing-resistant multifactor authentication (MFA) and passkeys, disciplined identity hygiene, tiered administration, and strong enforcement of privileged access and least-privilege principles. What is changing is the scope. The identity control plane now spans human and non-human identities, including applications and agents. Every AI agent or automated tool that signs in to your systems needs an owner, a defined level of access, and a review date, the same as any colleague with a login.
“Most organizations can tell you how many colleagues have accounts,” said Joshua Schmidt, Partner, Cyber and IT Advisory at BPM. “Far fewer can tell you how many service accounts, integrations, and AI tools have access to the same systems, who approved that access, or when anyone last reviewed it. That inventory is the first thing we ask for, and it is where the gaps tend to show up.”
Data Governance Determines What AI Can Reach
AI platforms are only as trustworthy as the data behind them, and Microsoft’s report treats data protection as a long-standing priority that has grown more urgent. In Microsoft’s words, “now that AI systems and agents can reach and act on sensitive information at unprecedented scale, it’s more crucial than ever.”
In practice, an AI tool or agent that can reach a file share can act on everything in it. Stale permissions, unlabeled sensitive records, and broad shared folders expose you to a new kind of mistake as well as a new kind of attacker. Classifying sensitive data, trimming access to what each role requires, and knowing where regulated information lives, such as client records, payroll files, and financial reporting, make AI adoption safer and make an incident easier to contain.
Questions to Answer Before Year End
Use these to test where your organization stands today:
- Who holds privileged access, and when was it last reviewed?
- Do administrators and finance colleagues use phishing-resistant MFA?
- Which AI tools and agents can reach your sensitive data, and who owns each one?
- How do you decide which vulnerabilities in internet-facing systems to fix first?
- Do you know where your sensitive data lives and who can access it?
Microsoft’s report also calls for protecting AI software supply chains, prioritizing the security of agentic AI systems, and evolving detection and response to match the pace of AI. Clear answers to the questions above put you in a stronger position to act on each of those.
Find the Gaps Before an Attacker Does
BPM’s cybersecurity team pairs hands-on Cybersecurity Assessments that map your identity, access, and data exposure with Managed IT services that keep the fixes in place, including access reviews, patching, monitoring, and documentation that holds up when a customer, an insurer, or a regulator asks how you manage risk.
If you are unsure how many identities, human and otherwise, can reach your most sensitive data, connect with BPM’s Cybersecurity team to find out before someone else does.
Linsey Gallo
Cybersecurity Compliance Manager, Advisory
Linsey helps organizations bridge the gap between business objectives, technology decisions, and cybersecurity risk through virtual Chief Information Security Officer …
Jon Parrish
Technical Assessment Manager
Jon Parrish supports BPM clients by performing security tests against their organization’s network. This can take many forms depending upon …
Josh Schmidt
Partner, Advisory
Josh started his career building IT systems in 2009 and has nearly a decade of experience working directly with clients …
Start the conversation
Looking for a team who understands where you’re headed and how to help you get there? Whether you’re building something new, managing growth or preserving success, let’s talk.