Big Blue Says AI Is Not Behind The Big Uptick In IBM i Security Alerts
November 16, 2026 Alex Woodie
IBM has issued nearly 400 security alerts for IBM i so far in 2026, a 5X increase compared to all of 2025. IBM says the big uptick in alerts and associated security patches is due to a general increase in security vulnerabilities across the industry, and is not due to its use of AI to help it find flaws and get patches into the hands of customers.
If you follow the IBM i PTF Guide on a weekly basis – which is highly recommended – you may have noticed more security alerts and patches this year. It turns out, it is a lot more, as an analysis of the security alerts that IBM provides on the IBM Product Security Central page of the IBM Support website shows.
The Product Security Central allows users to search by keyword. Using the keyword “IBM i” yields 1,312 results going back 12 years. The vast majority of these alerts for security flaws appearing in the Common Vulnerabilities and Exposures (CVE) database also contain links to IBM program temporary fixes (PTFs) to remediate the flaws. (For the vast majority of security alerts, IBM also issues emergency PTFs. However, it has issued security alerts without PTFs in 2026.)
The trend from 2020 to 2026 is telling. In 2020, IBM issued 55 security alerts for IBM i, followed by a year with just 38 alerts. In 2022, it issued 93 security alerts for IBM i, followed by two years with a nearly identical number, 68 and 69, respectively. The number of alerts bumped up to 77 in 2025, before skyrocketing to 396 alerts for CVEs with “IBM i” in the title over the first 10-and-a-half months of 2026, representing more than a 5x increase in the pace of alerts. Going back 12, years, more than 30 percent of all of the security alerts IBM has issued with “IBM i” in the title (a good but not perfect proxy for vulnerabilities impacting the server) are from 2026.

The number of alerts issued by IBM containing the keyword “IBM i” in the title of the security alert, per the IBM Product Security Central website. (Click image for link.)
There are a couple of caveats to this analysis. For starters, IBM doesn’t distribute data on IBM i vulnerabilities alone. It has thousands of products, and “IBM i” is apparently not one, which forces us to do a keyword search for alerts with “IBM i” in the title (which as mentioned is useful but not perfect).
The data is also a bit discombobulated in the early years. Nearly all of the patches for the period from 2014 to 2019–which covered 515 patches for CVEs involving IBM i – were inserted into the database by IBM over the course of just two days, December 18 and 19, in late 2019. (For some reason, IBM also listed two security PTFs for 2018 separately.)
This search also does not include security alerts for related components. For instance, it doesn’t include any security alerts for Rational Developer for i. According to the website, there are 279 alerts for RDi going back to 2014, and 99 of them are from 2026, meaning 35 percent of all alerts for RDi came in 2026. Db2 Mirror for i has 45 security alerts going back to 2022, 35 of which came in 2026 (a 78 percent rate). A search for Bob yields 75 results, all from 2026. IBM Rational Development Studio for i has just one security alert over the course of a dozen years: a local privilege escalation from 2024 that has a per the Common Vulnerability Scoring System (CVSS) base score of 8.4 (naturally, you should patch it if you haven’t already).
In terms of severity, the distribution looks pretty much like what you would expect. Of the 396 security alerts IBM issued for 2026 for IBM i-related security flaws appearing in the CVE database, 27 of the CVEs carry a low security rating per the CVSS base secure, 172 of the CVEs carry a medium security risk, 176 are high security risks, and 25 are for critical security risks.
IT Jungle asked IBM for an interview with the IBM i security architect to discuss the large increase in security alerts and how it impacts IBM i customers. IBM offered an email Q&A with an unnamed spokesperson instead. The unnamed IBM spokesperson said the increase is due to an increase in security vulnerabilities around the industry.
“IBM’s increase in security fixes is not unique. It is aligned with the general increase in security vulnerabilities across the industry,” the spokesperson said. “IBM has always been committed to timely and thorough security remediation, and that commitment will not change. Organizations should simply expect that any software or product they use, no matter the vendor, will have more updates. This is the new industry normal.”
Like other vendors, IBM is using AI tools to help find security flaws. In May, it joined Project Glasswing, a cross-industry cybersecurity initiative launched by Anthropic the month before. Anthropic launched Glasswing in in response to the concerns it had over the cyber hacking capabilities of its new large language model (LLM), dubbed Claude Mythos. Instead of releasing Mythos to the public, where bad actors could weaponize its expert-level AI hacking skills against unsuspecting victims, Anthropic decided to work with major system vendors to use Mythos to identify security flaws buried in source code.
Expert-level AI hacking tools were used to identify the critical PHP2Shell flaws that were disclosed last month. Ironically, the flaws were isolated at nearly the same time by two independent parties that were working with OpenAI’s GPT-5.6 Sol, which was that company’s version of an AI superhacker. Both PHP2Shell flaws had existed in the source code for more than a decade before AI discovered them.
The IBM spokesperson acknowledges that some of the increase in IBM i flaws and patches that IBM i customers are seeing this year are due to its use of AI. While it increases the volume, it doesn’t change the severity, the spokesperson said.
“IBM always is expanding our security tooling, and is doing so with Frontier AI technologies,” the spokesperson said. “These technologies do find more vulnerabilities that were missed by more traditional tooling, but the vulnerabilities are not special or unique. Such technologies simply find more of the same types of vulnerabilities. We evaluate the vulnerabilities found by all tooling, traditional and AI, together as a complete set, and remediate vulnerabilities based on severity and risk in context of the product and its typical use case. We do not prioritize vulnerabilities based on the source of how they were found.”
IBM is also using generative AI to help it fix the poorly written code at the heart of security flaws, as well as to generate the documentation that accompanies each PTF, including the PTF cover letters.
“Yes, IBM leverages code assistant tooling, such as IBM Bob, to help us accelerate the remediation of any identified vulnerabilities,” the spokesperson said.
Administrators have struggled under the flood of security alerts and emergency PTFs in 2026. The good news is IBM has been giving IBM i customers more tools to track CVEs. Let’s hope IBM keeps those improvements coming.

