- 91% of enterprise security leaders acknowledge their organizations deploy AI tools with direct access to internal systems.
- $4.63 million is the average breach cost for organizations with high levels of unapproved AI tool use.
- 127,000 WardenStealer infections recaptured since August 2026, targeting AI credentials.
Experts agree that the rise of AI agents and session hijacking attacks necessitates a fundamental shift in cybersecurity strategies, moving beyond traditional perimeter defenses to continuous identity monitoring and governance.
Beyond Passwords: How AI Agents and Session Hijacking Redefine Cyber Risk
AUSTIN, Texas – November 17, 2026 – The era of the brute-force password hack is rapidly drawing to a close. As organizations fortify their digital perimeters with multifactor authentication (MFA) and zero-trust architectures, cybercriminals are not retreating; they are evolving. According to a new 2027 forecast report released today by identity threat protection firm SpyCloud, the next wave of devastating cyberattacks will not break through the front door—they will walk right in using access that organizations already trust.
Based on insights from internal research teams and a survey of enterprise security professionals, the Austin-based company's report paints a sobering picture of the near future. The findings emphasize that attackers will increasingly bypass traditional security measures by targeting active employee sessions, vendor connections, and the rapidly proliferating ecosystem of artificial intelligence agents. As automated infostealers shift their focus toward API keys and SaaS integrations, the cybersecurity landscape is reaching a critical inflection point where standard perimeter defenses are rendered dangerously ineffective.
The Ghost in the Machine: AI Agents as Enterprise Blind Spots
The rapid integration of artificial intelligence into enterprise workflows has created an unprecedented security blind spot. Generative AI applications and autonomous coding assistants are no longer just tools; they act as synthetic employees. These non-human identities (NHIs) require extensive permissions, API keys, and session tokens to function, creating a sprawling, often unmonitored attack surface.
According to the report, 91 percent of enterprise security leaders acknowledge that their organizations deploy AI tools or autonomous agents with direct access to internal business systems. Yet, the governance of these tools severely lags behind their adoption. Industry research indicates that over 80 percent of employees use unapproved AI tools, contributing to a "shadow AI" crisis. Organizations with high levels of shadow AI experience average breach costs of $4.63 million, significantly higher than those with strict AI governance.
Cybercriminals have keenly observed this shift and are retooling their arsenals accordingly. Commodity infostealers, once designed primarily to scrape saved passwords from web browsers, are now explicitly engineered to hunt for AI credentials.
"With the proliferation of more SaaS and less on-prem services, NHIs have become as valuable if not more valuable than everyday traditional username/email password pairs," said Joe Roosen, Senior Director of Security Research at SpyCloud. "Because of this, commodity infostealers will pivot to seeking out NHIs as a preference."
Independent threat intelligence confirms this alarming trend. WardenStealer, a Rust-based infostealer marketed as Malware-as-a-Service, has recently been updated to target configuration files, access tokens, and prompt histories for AI tools like Claude, OpenAI's Codex, and GitHub Copilot. SpyCloud reports having recaptured more than 127,000 WardenStealer infections since August alone. Similarly, the fast-evolving Aoki infostealer has expanded its capabilities to siphon OAuth tokens and session files from a wide array of AI coding assistants. Because these stolen tokens often survive traditional password resets, they grant attackers persistent, undetected access to corporate networks and proprietary source code.
Beyond the Password: Why Session Hijacking is Bypassing MFA
For years, multifactor authentication has been the gold standard for enterprise security, a necessary friction point designed to thwart unauthorized access. However, the 2027 forecast highlights a technical evolution in cyberattacks that is effectively rendering legacy MFA implementations obsolete. Attackers are moving away from attempting new logins with stolen credentials and are instead focusing on hijacking active, already-authenticated sessions.
"The clearest pattern heading into 2027 is that attackers are gaining access through vectors that organizations already trust: active employee sessions, vendor connections, and even insider threats," said Aurora Johnson, Director of Threat Research & Intelligence at the firm. "Security measures like multifactor authentication are good, but defenders need to be mindful that they're not enough these days. Criminals have developed new tactics to effectively bypass these protections."
These new tactics include sophisticated Adversary-in-the-Middle (AiTM) and Browser-in-the-Middle (BitM) attacks. In an AiTM scenario, a malicious reverse proxy sits between the user and the legitimate login page. When the victim enters their credentials and completes the MFA prompt, the attacker intercepts the resulting session cookie. By injecting this cookie into their own browser, the attacker inherits the trusted session, completely bypassing the need for the user's physical device or authentication app.
Furthermore, the rise of device code phishing—a technique that abuses the OAuth 2.0 device authorization grant—is accelerating this trend. This method tricks users into authorizing an attacker-controlled application instance. Because the attack targets the authorization layer after a successful authentication, it can bypass even the most advanced phishing-resistant MFA methods, including passkeys. With over 30 distinct phishing kits now offering device code phishing capabilities, the barrier to entry for session hijacking has never been lower.
The Speed Gap: Automated Cybercrime and Supply Chain Vulnerabilities
As defenders scramble to secure non-human identities and patch MFA vulnerabilities, they are simultaneously facing an adversary whose operations are being supercharged by the very technology enterprises are struggling to govern. Artificial intelligence is widening the speed gap between attackers and defenders, allowing cybercriminals to automate mundane phishing campaigns, scale social engineering efforts, and rapidly stitch together synthetic identities using stolen personal data.
"AI is making familiar attacks faster and easier to scale," said Jason Lancaster, Chief Investigations Officer at SpyCloud. "Stolen sessions, API keys, and tokens already give criminals access to business systems, and as AI tools and agents hold more of those credentials, organizations need greater visibility into what has been exposed and the ability to remediate compromised access before attackers can exploit it."
This acceleration is particularly concerning when viewed through the lens of third-party risk. The modern enterprise is deeply interconnected, relying on a vast web of vendors, contractors, and SaaS providers. The forecast survey revealed that 68 percent of enterprise security professionals reported being impacted by a cyber breach originating from a third party. While this figure represents an acute observation among the surveyed group, it aligns with broader industry benchmarks, such as the Verizon Data Breach Investigations Report, which recently noted a 60 percent year-over-year increase in third-party breach involvement.
Attackers are increasingly leveraging long-forgotten vendor integrations—connections granted years ago that still hold administrative scope—to pivot into high-value targets. The automation provided by AI allows threat actors to map these interconnected supply chains with terrifying efficiency, identifying the weakest link in a vendor network and exploiting it to compromise downstream enterprise clients.
Navigating the New Perimeter
The convergence of AI-driven cybercrime, the proliferation of non-human identities, and the obsolescence of perimeter-based defenses demands a fundamental paradigm shift in enterprise security. The 2027 forecast serves as a stark warning that organizations can no longer rely on establishing trust solely at the point of login.
As threat actor groups continue to migrate their communications to feature-rich, invite-only chat applications like Session and Matrix to evade law enforcement, tracking their movements will become increasingly difficult. The focus for enterprise defenders must shift from merely preventing initial access to continuously monitoring the behavior of identities—both human and synthetic—within the network.
Security analysts stress that mitigating these emerging threats requires a comprehensive approach to identity-first security. This includes aggressive SaaS security posture management, the implementation of continuous authentication mechanisms that evaluate trust throughout the lifecycle of a session, and rigorous governance over the deployment of AI agents. Until organizations can achieve total visibility into the API keys, session tokens, and OAuth permissions floating within their environments, the ghost in the machine will continue to hold the keys to the kingdom.
Topics & Related
Threat Landscape
Identity & Access Management
📝 This article is still being updated
Are you a relevant expert who could contribute your opinion or insights to this article? We'd love to hear from you. We will give you full credit for your contribution.
Contribute Your Expertise →