IRONSCALES' third annual Cybersecurity Threat Report finds that 77% of organizations experienced an AI-agent-related security incident in the past 12 months [1], even as 96% of organizations now deploy AI agents with broad access to internal systems [1]. The report exposes a 'preparedness paradox': most organizations claim readiness on paper, yet only 52% have formal policies, controls, and incident response processes all in place simultaneously [1]. This gap represents a growing liability as the cybersecurity market expands toward $337.8 billion by 2029 [2].
What is Covered in this Article
- Agentic AI adoption outpacing security governance [1][1]
- AI-agent-related incidents already widespread at 77% of organizations [1]
- The preparedness paradox: confidence without controls [1]
- Deepfake defense ROI as a model for agentic AI urgency [1]
- IRONSCALES' platform positioning for agent-era email security [1]
The News: IRONSCALES released its third annual Cybersecurity Threat Report on October 21, 2026 [1][1], titled 'From Deepfakes to Double Agents: The New Face of AI-Driven Cyber Risk.' Conducted with Censuswide across 501 U.S. IT and cybersecurity professionals at organizations with 1,000 to 10,000 employees [1], the report finds that 77% of organizations say an AI agent or tool acted on malicious, misleading, or unauthorized content in the past 12 months, with more than half (57%) reporting a confirmed case [1]. AI agents are now deployed at 96% of organizations, with roughly four in five granting those agents access to internal systems [1]. Nearly half (49%) allow agent access to email inboxes or messaging platforms, and 30% grant access to finance, billing, or payment systems [1].
AI Agents Are Everywhere. Security Controls Are Not.
Analyst Take: The IRONSCALES report surfaces a structural vulnerability that most enterprise security programs have not yet addressed. AI agents have achieved near-universal deployment, but the governance frameworks that should accompany that deployment are conspicuously absent. With the cybersecurity market projected to grow at 11.6% CAGR through 2029 [2], the commercial pressure to solve this problem is intensifying alongside the threat.
Security Teams Sidelined as AI Agents Proliferate
The governance gap starts at the approval layer. IT holds primary authority over AI agents, plugins, and connectors at 78% of organizations, while security teams hold that role at just 6% [1]. This means the professionals best positioned to evaluate threat exposure are largely absent from deployment decisions. The consequences are already visible: among organizations that experienced an AI-agent-related incident, 39% report that an employee granted an AI tool access that security would not have approved [1]. With roughly four in five organizations giving agents access to internal systems [1], and 30% extending that access to finance, billing, or payment systems [1], the attack surface is substantial and growing faster than oversight.
Familiar Attack Vectors, New Targets
Attackers are not deploying novel techniques against AI agents. They are applying proven social engineering tactics to a new and largely undefended surface. Among organizations that experienced an AI-agent-related incident, the most common attack types were an AI tool summarizing or acting on a malicious email or message (42%) and an attacker using an AI tool during a phishing or social engineering attempt (41%) [1]. As IRONSCALES CEO Eyal Benishti noted, 'For years, the goal of a phishing email was to fool a person. Today it only has to fool the agent sitting between that person and their inbox.' That shift is registering with practitioners: 56% of respondents now view a deceived AI agent as at least as risky as a deceived employee [1].
The Preparedness Paradox: Confidence Without Controls
The report's most consequential finding is the gap between perceived and actual readiness. On paper, organizations appear well-prepared: 93% say their incident response plan covers AI agents, 93% say training covers safe AI use, and 84% are confident they could identify an AI-driven incident [1]. In practice, only 52% have formal policies, controls, and incident response processes all in place simultaneously [1]. This mirrors a broader pattern in the market: fewer than half of cybersecurity decision makers report being very confident in their organization's ability to detect a significant cybersecurity incident [3]. Confidence is not a control. The gap between self-reported readiness and documented process is where incidents happen.
Deepfake Defense Offers a Replicable Playbook
The report provides a concrete proof point for what focused investment can accomplish. The share of organizations actively investing in deepfake defense rose from 37% in 2025 to 48% in 2026 [1]. Among organizations that lost money to deepfake-related incidents, average losses fell approximately 20%, from roughly $280,000 to roughly $225,000, and seven-figure losses were nearly halved from 6% to 3% [1]. The trajectory is clear: urgency, followed by investment, followed by measurable loss reduction. The risk is that attention is already drifting. The share of organizations calling deepfake defense a top priority slipped from 71% to 60% as agentic AI competes for budget and focus [1]. IRONSCALES argues, credibly, that the same disciplined response now needs to be applied to agentic AI threats before losses compound.
IRONSCALES' Platform Positioning
IRONSCALES frames inbox security for AI agents as the logical extension of its existing email security platform, trusted by 18,000+ organizations and backed by a global threat intelligence network of 36,000+ security professionals across 3,500+ MSP partners [1]. The platform combines adaptive AI, agentic automation, and crowdsourced human intelligence to address phishing, business email compromise, deepfake, and GenAI-powered attacks. With nearly half of organizations already granting AI agents access to email inboxes and messaging platforms [1], and nearly half of cybersecurity decision makers planning budget increases in the next 12 months [3], the market timing for agent-aware email security is favorable. The question is whether enterprises act before the next wave of incidents forces the issue.
What to Watch
- Security governance adoption: whether organizations shift AI agent approval authority from IT to security teams in the next two quarters [1]
- Incident rate trajectory: whether the 77% AI-agent-related incident rate rises or stabilizes as formal controls catch up to deployment [1]
- Budget allocation signals: how cybersecurity decision makers planning modest increases [3] prioritize agentic AI controls versus other line items in Q1 2027 planning cycles
- Deepfake defense continuity: whether the priority-level decline from 71% to 60% [1] accelerates as agentic AI competes for the same budget and attention
- Competitive platform response: how incumbent email security vendors reposition or extend their offerings to address AI agent threat vectors over the next two quarters
Sources
1. New Report From IRONSCALES Reveals 77% of …, Ironscales, October 2026
2. 1H 2026 Cybersecurity Market Sizing & Five-Year Forecast, Futurum Research, June 2026
3. 1H 2026 Cybersecurity Global Enterprise Decision Maker Survey Report, Futurum Research, June 2026
Disclosure: Futurum is a research and advisory firm that engages or has engaged in research, analysis, and advisory services with many technology companies, including those mentioned in this article. The author does not hold any equity positions with any company mentioned in this article.
Read the full Futurum Group Disclosure.
Other Insights from Futurum:
How IRONSCALES' Recognition Signals a Shift in Email Security Dynamics
AI-Driven Phishing Defenses Increase Costs for Security Teams
NVIDIA Simulates Fault-Tolerant Quantum at 104 Qubits
Author Information
This content is written by a commercial general-purpose language model (LLM) along with the Futurum Intelligence Platform, and has not been curated or reviewed by editors. Due to the inherent limitations in using AI tools, please consider the probability of error. The accuracy, completeness, or timeliness of this content cannot be guaranteed. It is generated on the date indicated at the top of the page, based on the content available, and it may be automatically updated as new content becomes available. The content does not consider any other information or perform any independent analysis.
