Most Organizations Skip Permissions Reviews Before Deploying AI Tools

AI is being deployed faster than the data foundation beneath it is being checked, a new Syskit study has revealed.

In its latest State of Microsoft 365 Governance Report, published on September 10, security governance firm Syskit has found that three-quarters (76%) of organizations in the UK and the US  have deployed or piloted an enterprise AI tool such as Copilot on Microsoft 365 data.

Despite this wide adoption, only 43% of respondents confirmed they had completed a thorough review of permissions and oversharing risk before deploying these tools. The rest admitted to only having run a partial review or none at all.

Additionally, the study showed that controls for AI agents lag well behind the confidence in them: despite 91% of respondents claiming they are confident they can see which agents are active and what those agents can reach, only 22% confirmed having a formal policy defining what AI agents may access, while one in 10 (9%) let an agent inherit the full permissions of whoever deployed it.

Toni Frankola, CEO of Syskit, said that AI agents have removed the friction that once made accidental access to sensitive files less likely. Copilot and other AI tools can now surface content based on existing permissions, including files and sites that were shared broadly years ago and have not been reviewed since.

“What stands out in this data is that so few organizations can check what their AI can actually reach before switching it on, and fewer still plan to spend anything on finding out. Reviewing permissions is unglamorous work, but it has become the deciding factor in whether an AI rollout is safe,” he added.

Microsoft 365 Permissions Leave Organizations Exposed

Other findings show that misconfigurations and permission failures across Microsoft 365 are still prevalent in most organizations.

The risk is embedded in the permission model itself, with 41% of organizations leaving SharePoint sites accessible to all staff without restrictions, 35% of respondents admitting former employees’ files remain available to active users and 33% saying they have files shared with “Everyone.”

The leading governance challenge is content with no accountable owner. Around half (47%) identify orphaned teams, groups and sites as a key concern. Without active ownership, such content is less likely to be reviewed, removed or secured, yet AI tools can access it with the same authority as any other content.

Organizations also appear more confident in their access controls than they are able to evidence. While 83% say they know exactly who can access sensitive data at any given moment, only 4% could produce a complete access report for an external auditor within an hour. The majority (55%) would need a day or longer.

Overall, nine in 10 organizations (90%) have experienced, or suspect they have experienced, a security incident linked to misconfiguration or over-permissioned access in the past two years, and 39% have confirmed one.