Companies have been investing in AI systems for years, and most seem poised for the technology to have a measurable impact on their operations in 2026. But data security platform Kiteworks has found issues with the ways many companies’ AI systems are structured. In its new Data Security and Compliance Risk: 2026 Forecast Report, Kiteworks found many gaps in data frameworks that support AI.

Kiteworks Chief Strategy Officer Tim Freestone said that companies have spent time building governance frameworks that operate on paper—but they’d be hard pressed to actually show what’s happening with their data in practice. The study found that only about a third have visibility into where external partners process, train or use their data. A total of 57% don’t even have centralized data gateways to track data flows across their own environments.

Another major issue Kiteworks found in its study: Most enterprises haven’t truly established controls over AI agents. A total of 63% haven’t developed limitations on what AI agents are supposed to do, and six in 10 don’t have a “kill switch” in case an agent goes rogue. Less than half can isolate an AI agent to one part of their network or validate that the data it’s receiving is correct. The study found that at this point, many companies have invested in monitoring their agents, meaning they can see if they’re doing something wrong, but could have trouble stopping them.

Firms are working on closing these gaps to increase their internal and external security, especially as they continue to expand their AI systems. Kiteworks recommends that companies immediately look into limitations and kill switches for AI agents, work to audit and map data trails, figure out exactly how third parties are using their data. More actionable governance is needed—especially on the company board level, where more than half don’t have AI and security on their agenda—and enterprises need to be able to have more control over everything their data does.

As everyone in IT knows at this point, the best time to straighten out problems with new systems is as soon as they’re noticed. As time goes by, more systems will be built on top of today’s AI, and smaller problems in the foundation can blow up to existential threats over time.

Another constant problem that every IT leader should have on their radar is large-scale network outages. In the last several months, outages within AWS and Cloudflare crippled many systems businesses depend on for hours. While the big tech companies work toward being more resilient, every company also should take steps to keep vital functions running—or at least let customers know what is going on. I talked to Chris Angus, vice president of CPaaS and CX expansion for cloud communications and CX software provider 8x8, about how and why to do this. An excerpt from our conversation is later in this newsletter.

Leading AI companies are kicking off 2026 with new funding rounds and new business. Anthropic, the developer behind the Claude chatbot, is hoping to raise $10 billion in its next funding round, which would boost its market value to $350 billion—close to twice its current $183 billion valuation. According to the Wall Street Journal, this round is likely to be led by Singapore’s sovereign wealth fund GIC and Coatue Management. This would be the third huge funding round for Anthropic in 12 months, following a $13 billion round in September and a $3.5 billion round in March.

This would be the beginning of a massive year for Anthropic, which Reuters reported is projecting to nearly triple its annual revenue in 2026,........

© Forbes