Look into our cybersecurity crystal ball for the remainder of 2026, and also you in all probability will not be stunned to see a well-known acronym seem: AI.
What’s new this 12 months is that — three years after ChatGPT first burst into public consciousness — CISOs at the moment are getting right down to the nitty-gritty of AI threats and alternatives like by no means earlier than. Excessive-level considerations about AI provide chain threat have given strategy to granular discussions of audit-ready AI exercise logs, software-bill-of-materials (SBOM)-style mannequin attestation and Mannequin Context Protocol (MCP) server safety. Moreover, largely theoretical musings on AI brokers’ potential to remodel the safety operations middle (SOC) have now turn out to be sensible conversations about breaking defensive duties into agentic workloads.
You may discover all of this and extra within the following assortment of 2026 cybersecurity predictions, as shared with SearchSecurity by your fellow trade leaders.
1. AI-enabled social engineering campaigns will escalate
Suppose that is your boss on Zoom? Suppose once more. Many specialists predict 2026 would be the 12 months that typical enterprise customers study — some the onerous means — that they’ll now not belief their very own eyes and ears.
“We’re poised to see a brand new part of cyber-risk in 2026,” warned Andy Ulrich, CISO at Vonage, a part of Ericsson. That is as a result of attackers are utilizing generative AI and deepfakes to launch more and more convincing phishing assaults at scale, no matter native language and social engineering abilities.
Enterprises, Ulrich added, should double down on coaching customers to method each digital interplay — even with trusted colleagues — with wholesome skepticism. At Vonage, for instance, he has already begun together with AI-enabled social engineering eventualities in safety consciousness coaching to reveal what such assaults appear like in follow.
“It is turn out to be more and more essential for workers throughout departments to know what’s doable so they’re higher ready for the elevated stage of sophistication that AI will convey to the desk,” he mentioned.
2. Safety defenders will embrace agentic AI — or fall behind
The very best safety packages in 2026 will not essentially have the largest budgets, predicted Sergio Oliveira, director of improvement at DesignRush, a B2B design and advertising company listing platform. Somewhat, they would be the ones utilizing agentic AI to its full potential, moderately than merely treating it as a “shiny new toy.”
This 12 months, AI-enabled menace intelligence will turn out to be the linchpin of recent safety structure, Oliviera added. “Agentic AI will act as your analyst in actual time, ingesting and correlating alerts throughout identification, purposes, cloud and endpoints quicker than any present human groups can do.”
To appreciate the know-how’s potential within the SOC, mentioned GitLab CISO Josh Lemos, defenders should scale up their agentic AI instrument chain in the identical means adversaries break down assault phases into separate agentic workloads.
“AI brokers that leverage inner system sources — methods that present deep visibility into supply code, infrastructure code, software program composition and dependencies — can break down defensive duties to determine and remediate vulnerabilities previous to adversarial exploitation,” Lemos mentioned.
AI may show particularly useful for SMBs with restricted IT and safety employees, added Jason Ruger, CISO at PC maker Lenovo, particularly by augmenting their incident response capabilities. “That is one thing I am optimistic about,” he mentioned.
3. Boards and auditors will maintain CISOs accountable for AI safety
With nice energy comes nice accountability — and in 2026, AI is nothing if not highly effective. Some specialists imagine that govt stakeholders will more and more maintain CISOs accountable for AI governance and safety.
“Anticipate boards to demand audit-ready logs for each AI motion, plus SBOM-style attestation for fashions and information lineage,” mentioned Andrei Blaj, co-founder of medical imaging infrastructure supplier Medicai. “CISOs get ‘protected harbor’ provided that these controls are stay.”
To handle such considerations, Blaj mentioned Medicai plans to deploy a digital non-public cloud for AI, implement no-AI fallbacks and set up policy-as-code that permits directors to deactivate an AI agent with a single click on. “That is proof you may present a regulator at 2 a.m.,” Blaj added.
Safety leaders needs to be ready to supply AI audit trails, mannequin threat assessments and incident response readiness plans, mentioned DesignRush’s Oliveira. “Regulators will count on CISOs to reveal not simply an funding in know-how however the institution of efficient governance processes,” he mentioned. “‘I did not know’ will now not be a legit protection.”
4. AI regulation will turn out to be a serious problem for distributors
Lenovo’s Ruger mentioned he expects organizations like his that provide AI providers will wrestle to cope with tightening laws, particularly on the privateness entrance.
“Within the subsequent 12 months or two, we are going to see, like with the EU AI Act, laws that say an organization that gives AI providers wants to observe what’s being requested of the LLM and what the LLM is outputting,” Ruger mentioned.
He added that, as a personal firm, Lenovo’s place is to go away the choice to its prospects. If a buyer desires full privateness, Lenovo won’t ever know what they ask the LLM or what the mannequin outputs. Ruger famous, nevertheless, that regulators do not essentially appear thinking about permitting prospects the selection to decide out of getting suppliers monitor their use of LLMs.
5. MCP server threats would require new safety instruments and practices
MCP servers join AI fashions to third-party information, instruments and providers, underpinning many essential enterprise AI use instances. However the explosion in MCP server recognition additionally brings a bunch of recent safety dangers, cautioned Dave Stapleton, chief belief officer at ProcessUnity, a third-party threat administration service supplier.
“There are millions of MCP servers now on the market — many barely used or monitored — and it is clear the ecosystem is shifting quicker than the safety practices round it,” Stapleton mentioned. Dangers embrace weak entry controls, misconfigured permissions and software program provide chain points, in addition to novel AI threats reminiscent of command injection, instrument poisoning and context spoofing.
“We have to begin treating MCP like a essential integration layer with actual guardrails, visibility and accountability baked in,” he mentioned. “As MCP adoption continues to develop, we will count on a rise in purpose-built safety instruments to assist enterprises determine and handle related dangers.”
Stapleton mentioned he significantly hopes to see the emergence of instruments and providers that provide the next:
- MCP server code scanning.
- Expanded runtime monitoring of MCP server actions and connections.
- Integration of MCP servers with SIEM methods and different safety information aggregation instruments.
- MCP server gateways or proxies.
- MCP server threat assessments.
- The appliance of zero-trust ideas to MCP server know-how.
6. Safety capabilities will drive AI shopping for choices
Considerations about AI threats and vulnerabilities will more and more affect B2B buying choices in 2026, predicted Docusign CISO Michael Adams, including that inbuilt safety and compliance capabilities will turn out to be the last word AI tooling differentiators.
“We’ll see a shift towards platforms that may scale innovation safely, pairing AI-driven effectivity with the identical rigor historically reserved for essential infrastructure,” Adams added. “Probably the most profitable firms will probably be those who deal with belief as a design precept, guaranteeing each AI functionality meets enterprise-grade safety and compliance requirements.”
7. CISOs will tackle extra IT tasks
Emilio Escobar, CISO at observability and safety platform vendor Datadog, predicted that cybersecurity leaders will more and more tackle conventional CIO features, enabling them to personal IT moderately than simply report into IT.
“I see extra CISOs now being chargeable for IT or having some form of operational infrastructure accountability,” mentioned Escobar, who beforehand held safety positions at Hulu and PlayStation. At Datadog, for instance, safety owns each IT and — as of a few 12 months and a half in the past — web site reliability engineering.
“It simply made sense to turn out to be extra of a resiliency operate moderately than reliability and safety being seen as two separate parts of threat,” Escobar mentioned.
8. Passwords will turn out to be much less widespread
Passwords stay a stubbornly mounted ingredient of recent work life. In lots of workplaces, they’re as unloved as they’re unavoidable. Whereas rumblings of going passwordless have been occurring for years, it would lastly be the know-how’s time to shine.
Distant entry vendor TeamViewer has already taken the plunge, and the corporate’s CISO, Jan Bee, predicted that the advantages of passwordless will encourage extra firms to make the change.
“We enforced many new measures in our identities,” Bee mentioned, noting that implementing passkeys to bind a consumer’s identification to a selected gadget has not solely helped cut back digital friction, however has additionally added a stage of comfort that at the beginning appeared virtually too good to be true. “You do not have to recollect any password — there is no such thing as a password,” he mentioned.
To ease customers’ considerations about passwordless and passkey safety, Bee really useful explaining that biometric information is saved regionally on a tool, decreasing the chance of identification theft in a breach. He additionally pressured the significance of addressing vulnerabilities reminiscent of session theft as a secondary layer of protection.
Alissa Irei is senior web site editor of Informa TechTarget’s SearchSecurity.
Phil Sweeney is an trade editor and author targeted on cybersecurity matters.
