Listen to this Post
Introduction: The Hidden Risk Behind the AI Revolution
Artificial intelligence agents are rapidly transforming modern enterprises. From automating legal research and customer support to managing workflows and analyzing vast datasets, AI agents are becoming indispensable business tools. Organizations are racing to deploy these autonomous systems to increase productivity, reduce costs, and accelerate decision-making.
Yet beneath this wave of innovation lies a growing security challenge. As AI agents gain greater autonomy, they increasingly operate beyond the visibility of traditional security controls. Their actions can unfold in milliseconds, making it difficult for security teams to detect harmful behavior before damage occurs. This emerging threat landscape has given rise to a new generation of cybersecurity startups focused specifically on AI security.
One of the latest companies entering this space is Tenet Security, a startup founded by former Cisco AI Defense researchers who believe AI agents represent both the greatest productivity breakthrough and one of the most significant security challenges enterprises have ever faced.
A New Cybersecurity Startup Focused on AI Agents
Originating in Tel Aviv-Yafo, Israel, before establishing headquarters in the United States, Tenet Security has emerged from stealth mode with a mission centered on protecting organizations from dangerous AI agent behavior in real time.
The company was founded by Barak Sternberg, who serves as CEO, and Nevo Poran, the company’s CTO. Both founders share deep cybersecurity expertise and previously collaborated in building Cisco’s AI Defense research team. Their backgrounds also include service in Israel’s renowned Unit 8200 intelligence division, often regarded as a breeding ground for cybersecurity innovation.
The startup recently secured seed funding led by Westly Group, a venture capital firm known for backing notable technology companies including Tesla, SentinelOne, and Luminar. The investment reflects growing confidence that AI security will become one of the most important sectors within cybersecurity over the coming decade.
Why AI Agents Create a New Security Problem
Traditional cybersecurity tools were designed to monitor users, applications, devices, and networks. AI agents introduce an entirely different operational model.
Unlike conventional software, AI agents can independently make decisions, interact with systems, access data, call APIs, and collaborate with other AI systems. This autonomy creates behavior patterns that security tools often struggle to understand.
According to Tenet, once AI agents become embedded within enterprise environments, much of their activity effectively disappears from the security team’s view until after actions have already occurred.
This visibility gap creates a dangerous situation where organizations may unknowingly allow autonomous systems to perform sensitive operations without sufficient oversight.
As enterprises deploy increasing numbers of AI agents, security teams face the challenge of understanding what those agents are doing, why they are doing it, and whether their actions remain aligned with business objectives.
Understanding the Threat of Runaway AI Behavior
One of
A runaway agent is not necessarily malicious. Instead, it may begin executing actions that deviate from intended objectives due to flawed instructions, unexpected environmental conditions, logic errors, or unforeseen interactions with external systems.
Because AI agents can operate continuously and at machine speed, even minor errors can rapidly escalate into significant operational or financial consequences.
An AI agent configured to optimize a business process could accidentally consume excessive computing resources, generate unnecessary costs, or trigger unauthorized actions across interconnected systems before human operators notice anything unusual.
The challenge becomes even more severe when hundreds or thousands of AI agents are operating simultaneously across enterprise environments.
The Rise of Agentjacking Attacks
Beyond accidental behavior, Tenet highlights a more dangerous threat category known as “agentjacking.”
Agentjacking occurs when attackers manipulate AI agents through poisoned data, malicious inputs, compromised information sources, or carefully crafted interactions that alter agent behavior.
Unlike traditional cyberattacks that rely on exploiting software vulnerabilities, agentjacking targets the decision-making process itself.
Attackers may attempt to influence an AI agent into exposing sensitive information, escalating privileges, accessing restricted resources, or performing actions that ultimately benefit malicious actors.
According to Tenet, these attacks are particularly difficult to detect because the AI agent may technically be operating within its authorized permissions while still carrying out harmful activities.
This creates a scenario where traditional security systems see no obvious violation even as the organization faces increasing risk.
How
To address these challenges, Tenet has developed a patent-pending security platform specifically designed for autonomous AI environments.
The
These sensors observe operating system behavior, network communications, API interactions, and even the reasoning processes generated by large language models.
By combining these visibility layers, the platform attempts to create a comprehensive understanding of how AI agents behave in real time.
Tenet believes many organizations currently underestimate the number of active AI agents operating within their environments. In some cases, the actual number may be several times higher than security teams realize.
The
Predicting Dangerous Actions Before They Occur
Perhaps the most distinctive feature of
When the platform identifies suspicious activity, it does not merely generate an alert.
Instead, it attempts to simulate and forecast the AI agent’s next actions before execution occurs.
If the projected behavior appears harmful, the platform can intervene and block the action before damage is done.
This represents a shift from traditional reactive cybersecurity toward proactive prevention.
Rather than investigating incidents after they happen, organizations can potentially stop dangerous AI actions at the exact moment decision-making occurs.
For autonomous systems operating at machine speed, such predictive protection could become essential.
Research Findings Raise Industry Concerns
Tenet’s internal Threat Labs have reportedly conducted extensive research into agentjacking techniques across enterprise environments.
According to the company, researchers validated attack methods across more than one hundred enterprise environments and identified thousands of organizations potentially exposed through publicly accessible attack paths.
The findings suggest that AI agents may already be introducing security risks that many enterprises have not yet fully recognized.
Because these attacks leverage legitimate permissions and trusted workflows, they often bypass conventional security controls designed to detect known malicious activity.
As AI adoption accelerates, such vulnerabilities could become increasingly attractive targets for sophisticated threat actors.
Early Customer Deployments Show Promising Results
Tenet reports positive outcomes from several early enterprise deployments.
One legal-sector organization generating approximately $1 billion in annual recurring revenue reportedly expanded its AI agent deployments from just two implementations to more than twenty within six months while using Tenet’s platform.
During that period, the company claims more than ten attack attempts were detected and blocked, including a critical cross-site scripting attack.
In another Fortune 1000 deployment, Tenet reportedly identified a runaway AI agent responsible for generating tens of thousands of dollars in unnecessary token consumption over a single weekend.
By detecting the issue early, the organization was able to prevent broader financial impact.
While these examples remain company-reported results, they highlight the growing demand for visibility into autonomous AI systems.
The Future of AI Security
The emergence of companies like Tenet reflects a broader transformation occurring across cybersecurity.
For decades, security strategies focused on protecting humans from attackers. The next era may focus on protecting organizations from unintended or manipulated actions performed by autonomous systems acting on behalf of humans.
As AI agents become increasingly integrated into enterprise operations, security solutions capable of understanding machine decision-making will likely become as essential as traditional endpoint protection and network monitoring.
Organizations that successfully balance innovation with security may gain significant competitive advantages, while those that overlook AI-specific risks could face new categories of operational and financial exposure.
Deep Analysis: Runtime Security Through a Linux Lens
The concept behind
Monitoring agent behavior resembles process monitoring:
ps aux top htop
Observing network communications aligns with:
netstat -tulnp ss -tulnp tcpdump
Investigating suspicious system activity reflects:
auditctl ausearch journalctl
Tracking API and application behavior resembles:
strace ltrace
Monitoring file system changes can be achieved with:
inotifywait auditd
Detecting privilege escalation attempts aligns with:
sudo -l last who
Analyzing runtime execution behavior reflects:
perf systemd-cgtop
Threat hunting activities commonly use:
grep awk sed find
Containerized AI environments can be monitored through:
docker ps docker logs kubectl logs
The underlying philosophy remains consistent. Visibility at runtime provides the earliest opportunity to identify dangerous behavior before significant damage occurs. Tenet effectively applies this traditional cybersecurity principle to the emerging world of autonomous AI agents, extending runtime monitoring beyond humans and applications into machine decision-making itself.
What Undercode Say:
The cybersecurity industry is currently witnessing the birth of an entirely new security category. For years, organizations focused on protecting infrastructure, endpoints, cloud assets, identities, and applications. AI agents now introduce a sixth major security layer.
What makes Tenet particularly interesting is not simply its detection capabilities but its focus on behavioral understanding.
Traditional security products generally evaluate events after they happen.
AI agents operate too quickly for that model.
An autonomous system can perform thousands of actions in the time it takes a human analyst to review a single alert.
The
Many organizations have deployed AI tools faster than they have developed governance frameworks.
This creates a dangerous gap between innovation and oversight.
The concept of agentjacking is especially noteworthy.
Cybercriminals have historically targeted software vulnerabilities.
The next generation of attacks may focus on influencing decision-making engines rather than exploiting code execution flaws.
This represents a fundamental shift in attacker strategy.
Prompt manipulation, data poisoning, memory corruption, and contextual deception may eventually become as common as phishing and malware.
If that occurs, conventional security platforms will struggle to respond effectively.
Another important aspect is economic risk.
Not every AI incident results in data theft.
Runaway agents can generate significant operational expenses through unnecessary API calls, token consumption, cloud resource allocation, and automated transactions.
Financial damage alone could justify dedicated AI security investments.
Tenet’s reported discovery of excessive token spending demonstrates that AI security is not purely a defensive discipline.
It is also becoming an operational cost management function.
The
Building
Unit 8200 alumni have historically played influential roles in cybersecurity innovation.
Investor interest is equally telling.
Venture capital firms rarely invest heavily in categories they believe will remain niche.
The participation of a prominent technology investor signals expectations of substantial market demand.
The larger trend is impossible to ignore.
Every major enterprise software provider is introducing AI agents.
Microsoft, Google, Salesforce, ServiceNow, Oracle, and many others are building increasingly autonomous platforms.
As adoption accelerates, visibility challenges will expand proportionally.
Security vendors that understand AI reasoning, intent, and behavior will likely define the next generation of enterprise protection.
Tenet appears positioned to compete in that emerging landscape.
Whether it becomes a market leader remains uncertain.
However, its focus addresses a genuine and growing problem.
The industry is rapidly approaching a point where monitoring human users alone will no longer be sufficient.
Organizations must also monitor the machines acting on their behalf.
That reality may ultimately define the future of cybersecurity.
✅ Tenet Security was founded by former Cisco AI Defense team members Barak Sternberg and Nevo Poran.
✅ The company focuses on runtime detection and prevention of dangerous AI agent behavior before execution.
✅ Seed funding was led by Westly Group, a firm known for investments in major technology companies.
❌ Independent public validation of all reported deployment success metrics remains limited, as many performance figures originate from company statements rather than third-party audits.
❌ The long-term effectiveness of agentjacking prevention technologies has not yet been proven across large-scale global enterprise deployments.
❌ The AI security market is still emerging, meaning competitive and technological outcomes remain uncertain.
Prediction
(+1) Enterprise adoption of autonomous AI agents will accelerate significantly over the next three years.
(+1) Runtime AI security platforms will become a standard component of enterprise cybersecurity stacks.
(+1) Agent behavior monitoring will evolve into a dedicated cybersecurity discipline with specialized tools and analysts.
(-1) AI-powered attacks targeting agent decision-making will become more sophisticated and harder to detect.
(-1) Organizations that deploy AI agents without governance frameworks will experience rising operational and security incidents.
(-1) Traditional security solutions that lack AI behavioral visibility may struggle to protect future enterprise environments effectively.
▶️ Related Video (86% Match):
🕵️📝Let’s dive deep and fact‑check.
🎓 Live Courses & Certifications:
Join Undercode Academy for Verified Certifications
🚀 Request a Custom Project:
Secure, high-velocity infrastructure and disruptive technological engineering. Contact our engineering team for high-tier development and proprietary systems:
[email protected]
💎 Smart Architecture | 🛡️ Secure by Design | ⭐ Trusted by Thousands
References:
Reported By: www.securityweek.com
Extra Source Hub (Possible Sources for article):
https://www.linkedin.com
Wikipedia
OpenAi & Undercode AI
Image Source:
Unsplash
Undercode AI DI v2
🔐JOIN OUR CYBER WORLD [ CVE News • HackMonitor • UndercodeNews ]
📢 Follow UndercodeNews & Stay Tuned:
𝕏 formerly Twitter 🐦 | @ Threads | 🔗 Linkedin | 🦋BlueSky | 🐘Mastodon | 📺Youtube
