Artificial intelligence is turning cybersecurity into an arms race, and attackers are getting disturbingly creative with their methods. This guide is for cybersecurity professionals, business executives, and IT leaders who need to understand how AI is reshaping the threat landscape—both as a powerful defense tool and a dangerous weapon in criminal hands.

You’ll discover how cybercriminals are using AI to supercharge their attacks through sophisticated phishing campaigns that can clone voices and faces with scary accuracy. We’ll also explore the dark reality of adaptive malware that learns and evolves in real-time, making traditional security measures almost obsolete. Plus, you’ll see how AI is automating attack creation, allowing even amateur criminals to launch professional-grade cyber operations.

The cybersecurity battlefield has changed forever, and the weapons being deployed on both sides are more advanced than ever before. Get ready to learn why your current defenses might not be enough.

How AI Strengthens Cybersecurity Defense Systems

How AI Strengthens Cybersecurity Defense Systems

Automated Threat Detection and Real-Time Response Capabilities

AI-powered cybersecurity solutions have revolutionized threat detection by enabling organizations to identify and respond to attacks in real-time. Unlike traditional security systems that rely on pre-defined signatures and indicators of compromise, AI systems continuously learn from incoming data to detect even the most sophisticated threats.

Security Information and Event Management (SIEM) solutions enhanced with AI capabilities analyze security logs and events from various sources across an organization’s network. This enables significantly faster threat detection, investigation, and response times compared to manual processes. AI-driven endpoint security solutions proactively monitor laptops, desktops, and mobile devices, providing immediate protection against malware, ransomware, and other advanced persistent threats.

Network Detection and Response (NDR) solutions with AI capabilities continuously monitor network traffic patterns, identifying sophisticated threats that may bypass traditional security measures. These systems can automatically quarantine suspicious activities and initiate response protocols without human intervention, dramatically reducing the time between threat detection and containment.

Advanced Behavioral Analytics for Zero-Day Attack Prevention

Now that we have covered automated detection capabilities, behavioral analytics represents one of AI’s most powerful applications in cybersecurity defense. User and Entity Behavior Analytics (UEBA) systems analyze the activity patterns of devices, servers, and users to establish baseline behaviors for every entity within the network.

AI models develop comprehensive profiles of applications deployed on organizational networks and process vast volumes of device and user data. When incoming data deviates from established behavioral patterns, the system can immediately flag potentially malicious activity, even if it represents a previously unknown zero-day attack.

This approach is particularly effective because it doesn’t rely on known attack signatures or indicators of compromise. Instead, AI systems can protect businesses against vulnerabilities they are unaware of before they are officially reported and patched. The behavioral analytics approach enables organizations to identify evolving threats and unknown vulnerabilities that traditional signature-based systems would miss entirely.

Predictive Intelligence and Proactive Risk Assessment

With this behavioral foundation in mind, generative AI takes cybersecurity defense to the next level through predictive intelligence capabilities. By analyzing vast datasets of past attacks and security incidents, generative AI identifies patterns and trends that enable it to predict potential future attack scenarios.

This predictive capability allows organizations to stay one step ahead of cybercriminals and proactively implement countermeasures before attacks occur. Generative AI creates highly realistic simulations of cyberattacks, allowing security teams to test their defenses and incident response plans against a wide range of potential threats. This proactive approach helps identify vulnerabilities and improve preparedness before real attacks materialize.

Organizations can implement these predictive models to enhance their threat-hunting processes, moving from reactive to proactive security postures. The AI systems continuously analyze network traffic patterns over time, learning to recommend appropriate policies and security measures based on predicted threat landscapes.

Enhanced Brand Protection Through AI-Powered Monitoring

Previously, I’ve discussed how AI strengthens internal defenses, but AI also provides crucial brand protection capabilities through sophisticated monitoring systems. AI-powered phishing detection and prevention systems analyze the content and context of emails to quickly determine whether they are legitimate communications or part of malicious campaigns.

These systems can identify various indicators of phishing attempts, including email spoofing, forged senders, and misspelled domain names that target an organization’s brand reputation. Machine learning algorithms enable AI to learn from data patterns, making analysis increasingly accurate while evolving to address new threats targeting brand integrity.

AI tools integrated into authentication systems, including CAPTCHA, facial recognition, and fingerprint scanners, automatically detect whether login attempts are genuine or represent credential stuffing attacks targeting organizational accounts. Next-Generation Firewalls (NGFWs) infused with AI capabilities offer advanced threat protection and intrusion prevention, providing comprehensive brand protection across all digital touchpoints.

AI-driven cloud security solutions utilize artificial intelligence to protect organizational data and applications in cloud environments, ensuring both security and compliance while maintaining brand trust and reputation.

The Dark Side: How Cybercriminals Weaponize AI Technology

The Dark Side: How Cybercriminals Weaponize AI Technology

Now that we have covered how AI strengthens cybersecurity defense systems, it’s crucial to examine the darker implications of this technology. Cybercriminals are increasingly leveraging artificial intelligence to create more sophisticated, targeted, and challenging-to-detect attacks. These AI-powered cyberattacks represent a fundamental shift in the threat landscape, utilizing machine learning algorithms to automate, accelerate, and enhance various phases of malicious activities.

AI-Generated Phishing and Sophisticated Social Engineering

AI-driven phishing attacks have revolutionized the way cybercriminals craft their deceptive communications. Previously, creating convincing phishing emails required considerable sophistication and resources. Now, AI-powered tools like ChatGPT enable threat actors to quickly generate highly personalized and realistic emails, SMS messages, and social media outreach at unprecedented scale.

The sophistication of these attacks lies in their customization capabilities. AI algorithms excel at data scraping from public sources such as social media sites and corporate websites, gathering and analyzing information to create hyper-personalized, relevant, and timely messages. This enables attackers to:

• Identify ideal targets within organizations, focusing on individuals with access to sensitive data or those with lower technological aptitude
• Develop realistic personas and corresponding online presence for sustained communication
• Create plausible scenarios that generate attention and prompt action
• Write personalized messages that are nearly indistinguishable from legitimate communications

AI-powered chatbots have taken this threat to the next level by enabling real-time, automated interactions. These sophisticated chatbots can engage victims in casual conversations, often posing as customer support or service agents, to gather personal details, login credentials, and even reset account passwords. The ability to deploy these tools at scale allows attackers to attempt connections with countless individuals simultaneously.

Deepfake Technology for Executive Impersonation Scams

Deepfake technology represents one of the most concerning developments in AI-powered cybercrime. These AI-generated audio files, images, and videos are designed to deceive targets by impersonating trusted individuals, particularly corporate executives and leaders.

In executive impersonation scams, attackers use existing footage or audio recordings of corporate leaders to create doctored content that mimics their voice, mannerisms, and communication style. The sophistication of these deepfakes has reached a point where they can instruct employees to take specific actions such as:

• Authorizing urgent wire transfers
• Changing critical passwords or security settings
• Granting unauthorized system access
• Sharing sensitive corporate information

The effectiveness of deepfake voice scams has been demonstrated in real-world scenarios where employees, believing they were receiving direct instructions from their CEO, have transferred substantial funds to attacker-controlled accounts. These attacks exploit the natural human tendency to comply with authority figures, especially when the communication appears authentic and urgent.

Adaptive Malware That Evolves to Evade Detection

AI-enabled ransomware and malware represent a significant evolution in malicious software capabilities. Unlike traditional malware that follows predetermined patterns, AI-powered variants can learn and adapt in real time, making them increasingly difficult to detect and neutralize.

These adaptive malware variants leverage reinforcement learning to continuously evolve their techniques. Key characteristics include:

• Real-time adaptation: The malware learns from its environment and modifies its behavior to avoid detection by security systems
• Pattern avoidance: AI algorithms help the malware create attack patterns that security systems cannot easily identify
• Vulnerability exploitation: AI can automatically research targets and identify system vulnerabilities more efficiently than manual methods
• Encryption enhancement: AI assists in developing more sophisticated encryption methods for ransomware attacks

The self-modifying nature of these threats means that traditional signature-based detection methods become less effective over time, as the malware continuously alters its code structure and behavior patterns.

Automated Attack Creation Using Large Language Models

Large language models have become powerful tools for automating the creation of various attack vectors. Malicious GPTs – altered versions of generative pre-trained transformers – can produce harmful or deliberately misinformed outputs specifically designed to advance cyberattacks.

These AI systems can generate:

• Malware code: Automatically creating sophisticated malicious software with minimal human intervention
• Attack vectors: Identifying and developing new methods of system infiltration
• Fraudulent content: Producing fake emails, documents, and online content that support social engineering campaigns
• Vulnerability research: Accelerating the discovery of software weaknesses and system flaws

The automation capabilities extend to adversarial AI attacks, which specifically target AI and machine learning systems through:

• Poisoning attacks: Injecting fake or misleading information into training datasets to compromise model accuracy
• Evasion attacks: Applying subtle changes to input data to cause misclassification and negatively impact predictive capabilities
• Model tampering: Making unauthorized alterations to pre-trained models to compromise their output accuracy

With this understanding of how cybercriminals weaponize AI technology, it becomes clear that organizations must adopt equally sophisticated defensive measures to combat these evolving threats.

Critical Challenges and Limitations of AI in Cybersecurity

Critical Challenges and Limitations of AI in Cybersecurity

Alert Fatigue and False Positive Management Issues

AI systems in cybersecurity generate significant challenges when it comes to alert management and accuracy. While AI can process enormous volumes of data and identify potential threats at unprecedented speeds, this capability often comes with a critical drawback: the generation of false positives that overwhelm security teams. The reliability and trust issues surrounding AI systems stem from their tendency to make mistakes, creating concerns about their decision-making accuracy.

Security professionals must contend with AI systems that can miss genuine threats while simultaneously flagging legitimate activities as suspicious. This creates a dangerous scenario where teams become desensitized to alerts, potentially ignoring critical warnings buried within numerous false alarms. The challenge is compounded by the fact that AI systems’ decision-making processes aren’t always transparent, making it difficult for security teams to understand or predict their behavior patterns.

Algorithmic Bias and Black Box Decision-Making Problems

One of the most significant challenges facing AI implementation in cybersecurity is the persistent issue of algorithmic bias. AI systems are fundamentally limited by the quality and representativeness of their training data – if the data contains biases or is incomplete, the AI system will inevitably produce biased results. This creates particularly problematic scenarios in areas such as facial recognition and threat assessment, where bias can lead to false identifications and discriminatory outcomes.

The black box nature of many AI algorithms exacerbates these concerns. Security teams often cannot understand how AI systems arrive at their conclusions, making it nearly impossible to identify when bias is influencing critical security decisions. This lack of transparency raises serious questions about accountability and fairness in cybersecurity practices, particularly when AI systems make high-stakes decisions without adequate human oversight.

Cybersecurity Skills Gap and AI Expertise Requirements

The integration of AI into cybersecurity operations demands specialized technical expertise that many organizations currently lack. The complexity of combining AI technologies with existing cybersecurity infrastructure requires professionals who understand both domains thoroughly. This integration challenge is particularly acute when dealing with legacy systems, where compatibility issues demand significant technical knowledge and careful planning.

Organizations face the dual challenge of finding professionals who can manage AI tools effectively while also understanding the evolving landscape of AI-powered threats. The need for continuous education and training becomes critical as AI technology advances rapidly, requiring security teams to constantly update their skills to stay effective against sophisticated AI-driven attacks.

Privacy Concerns and Ethical Implementation Challenges

AI’s extensive data collection and processing capabilities create substantial privacy risks that organizations must carefully navigate. The potential for accessing or processing sensitive information without proper authorization poses significant threats to individual privacy rights. AI systems’ ability to analyze vast amounts of personally identifiable information raises concerns about data misuse and privacy violations.

The ethical implications extend beyond privacy to encompass broader questions about discrimination and equity in cybersecurity practices. Organizations must establish robust ethical guidelines to govern AI development and deployment, focusing on fairness, non-discrimination, and respect for privacy. The challenge is compounded by regulatory frameworks that often lag behind AI technological advancement, making compliance a moving target for organizations attempting to implement AI responsibly.

Real-World Applications and Success Stories

Real-World Applications and Success Stories

Case Studies of AI Preventing Ransomware Attacks

Now that we have explored the dark side of AI weaponization in cybersecurity, it’s crucial to examine how AI serves as a powerful defense mechanism against sophisticated attacks. Darktrace’s AI-driven platform stands as a prime example of successful ransomware prevention. At a global healthcare organization, Darktrace’s machine learning algorithms detected and responded to a ransomware attack in real-time, preventing the encryption of critical patient data. The AI system learned the normal behavior patterns of users, devices, and network traffic, enabling it to identify the suspicious data movement characteristic of ransomware deployment before any damage occurred.

Similarly, Boardriders faced a significant ransomware threat in 2021 that could have crippled their global retail operations across 700 locations. Darktrace’s Self-Learning AI was the first to respond, detecting the attack within minutes and alerting the security team through their mobile application. The autonomous response capabilities contained the threat independently, providing the small security team with critical hours to implement additional protective measures while maintaining business continuity across their 20 e-commerce sites and multiple warehouses worldwide.

Behavioral Baseline Detection in Enterprise Environments

With the increasing complexity of enterprise networks, behavioral baseline detection has become a cornerstone of modern AI cybersecurity strategies. Darktrace’s implementation at Boardriders demonstrates how AI establishes normal ‘patterns of life’ for every user and device within the organization. By continuously learning typical behaviors, the system can detect subtle deviations that may indicate account compromise or malicious intent.

IBM Watson for Cyber Security enhances this approach by analyzing user traffic patterns and communication behaviors to identify potential insider threats. The system processes millions of cybersecurity documents and correlates this information with internal behavioral data to establish comprehensive baselines. When Watson detects anomalies in user behavior – such as unusual data access patterns or communication with suspicious external entities – it provides actionable intelligence that enables security teams to investigate and respond proactively.

User and Entity Behavior Analytics (UEBA) powered by AI systems evaluate normal user activities and flag deviations that may signify compromised accounts. This capability proved essential for a global financial services firm that used IBM Watson to identify a sophisticated phishing campaign by correlating various behavioral data points across their enterprise environment.

AI-Powered Endpoint Protection Against Advanced Threats

Previously, we’ve seen how cybercriminals exploit AI for malicious purposes, but AI-powered endpoint protection represents one of the most effective countermeasures against these evolving threats. Cylance, now part of BlackBerry, revolutionized endpoint security by moving beyond traditional signature-based detection to predictive threat prevention. Their AI engine analyzes file and application characteristics before execution, utilizing machine learning models trained on billions of data points to identify both known and zero-day threats with exceptional accuracy.

A large manufacturing company successfully deployed Cylance to protect their industrial control systems (ICS) from targeted malware attacks. The AI’s pre-execution analysis capability blocked malicious files before they could execute, preventing potential disruption to critical production lines. This proactive approach proved essential in securing operational technology environments where traditional antivirus solutions often fall short.

Microsoft’s AI-powered security solutions demonstrate scalability in endpoint protection, processing over 6.5 trillion signals daily across their extensive product ecosystem. Their machine learning algorithms have improved malware and phishing detection rates by 40%, while reducing average threat detection time from 24 hours to under an hour. This rapid response capability is crucial for protecting endpoints in today’s fast-paced threat landscape.

Threat Intelligence Correlation Across Multiple Data Sources

The complexity of modern cyber threats requires sophisticated correlation capabilities that only AI can provide effectively. IBM Watson for Cyber Security exemplifies this approach by analyzing vast amounts of unstructured data from blogs, research papers, news articles, and threat intelligence feeds. The system’s natural language processing capabilities enable it to extract actionable intelligence from diverse sources, correlating this information with internal security data to identify emerging threats.

Microsoft’s Intelligent Security Graph processes an enormous volume of signals from multiple products and services, using advanced analytics to correlate threat indicators across different data sources. This comprehensive approach enables the identification of attack patterns that might remain hidden when examining individual data streams in isolation. The system’s ability to correlate seemingly unrelated events has resulted in a 60% decrease in successful cyber attacks on Microsoft’s infrastructure.

Darktrace’s approach to threat intelligence correlation focuses on combining external threat feeds with internal network behavior analysis. By correlating global threat intelligence with local network anomalies, the system provides context-aware threat detection that reduces false positives while improving the accuracy of threat identification. This multi-source correlation capability proved crucial for Boardriders, where Darktrace extended protection to their cloud environment, including Microsoft 365, providing comprehensive visibility into account takeovers and malicious activities across both on-premises and cloud applications during the shift to remote work arrangements.

Emerging Trends and Future Battlefield Dynamics

Emerging Trends and Future Battlefield Dynamics

Generative AI’s Dual-Use Nature in Offense and Defense

The emergence of generative AI has fundamentally shifted the cybersecurity landscape, creating unprecedented opportunities for both defenders and attackers. Security teams are increasingly leveraging generative AI to develop predictive models and enhance their capabilities to identify malicious activity quickly while automating responses to reduce threat response times. These AI systems can analyze massive datasets for anomalies, provide predictive threat intelligence, and automatically orchestrate rapid incident response procedures.

However, this same technology has become a double-edged sword. Cybercriminals are exploiting generative AI for sophisticated phishing attacks, creating deepfakes, and developing more convincing social engineering tactics. The technology enables threat actors to generate WormGPT-style attacks and create highly personalized phishing campaigns that are increasingly difficult to detect through traditional methods. This dual-use nature demands that security professionals understand both the defensive capabilities and offensive potential of generative AI technologies.

Regulatory Framework Development and Compliance Requirements

As artificial intelligence becomes more integral to cybersecurity operations, regulatory compliance has emerged as a critical consideration. The ethical implications of AI in cybersecurity encompass adhering to sensitive data protection laws, ensuring algorithm fairness, and maintaining transparency in automated decision-making processes. Organizations must navigate complex privacy considerations while implementing AI-powered security solutions.

The regulatory landscape is evolving to address the potential misuse of AI technologies, ensuring non-discrimination in threat detection, and maintaining accountability for AI-driven decisions and actions. Security leaders must balance the effectiveness of AI-powered threat detection with compliance requirements that protect user privacy and prevent algorithmic bias. This includes ensuring unbiased data use, protecting sensitive information, maintaining transparency in AI operations, and upholding user consent principles.

Autonomous Security Operations Centers and Self-Driving Defense

The future of cybersecurity operations is moving toward autonomous threat detection systems capable of identifying and mitigating cyber threats in real-time without human intervention. These advanced AI systems can continuously monitor network traffic patterns, identify previously impossible anomalies, and automatically execute response protocols to contain threats before they can cause significant damage.

Neural networks are being deployed to analyze complex data from thousands of distributed systems, identifying subtle anomalies that could indicate compromise. Machine learning algorithms are being trained to secure transactions by analyzing data patterns to detect fraud, while AI-powered systems provide automated vulnerability detection and adapt to mitigate new and evolving risks that cybercriminals continuously generate. This level of automation significantly reduces response time and manual workload while enhancing overall security posture.

The Arms Race Between AI-Powered Attackers and Defenders

The cybersecurity landscape has evolved into an intense arms race where both attackers and defenders leverage increasingly sophisticated AI technologies. Security teams are using machine learning and AI tools for early identification of sophisticated cyber threats, while cybercriminals employ AI for adversarial tactics to circumvent traditional security measures.

This competitive dynamic drives continuous innovation on both sides. Defenders are implementing enhanced phishing detection systems that expand on the use of Large Language Models (LLMs) to proactively identify and block malicious emails, while attackers develop more sophisticated methods to bypass these defenses. The identification of deepfakes using generative AI, LLMs, machine learning, and other AI tools represents just one battleground in this ongoing technological conflict.

The future battlefield will be characterized by continuous improvement in AI models for identity management and access control, extended use of AI-driven biometric and behavioral analysis for secure access control, and advanced automation and orchestration to streamline incident response. Organizations must remain vigilant and adaptive, as cyber threats constantly evolve, requiring AI systems that can learn and change based on new data and emerging attack vectors.

Strategic Implementation Guidelines for Organizations

Strategic Implementation Guidelines for Organizations

Now that we have covered the various ways AI impacts cybersecurity, organizations need concrete strategies to effectively implement AI-powered security solutions while navigating the complex challenges they present.

Building AI-Human Collaboration in Security Operations

Effective human-AI collaboration leverages the strengths of both AI and human expertise in cybersecurity operations. AI excels at processing large volumes of data and identifying patterns, but human oversight remains crucial for contextual understanding and decision-making.

The key is implementing AI as an assistant rather than a replacement, enabling true human-machine collaboration. Security professionals can focus on strategic tasks while AI handles routine monitoring and analysis. This approach empowers systems with the ability to learn without explicitly being programmed, allowing AI to find patterns within data to predict outcomes that were never seen before, enabling cybersecurity programs to scale without constant human intervention.

Regular training and feedback loops between AI systems and human operators continually improve AI performance. Organizations should establish clear protocols for when human intervention is required, particularly given that AI cybersecurity tools can generate false positives where benign activities are incorrectly flagged as malicious. These false alerts require human verification and resolution, making the human element essential for maintaining system accuracy and preventing alert fatigue among security professionals.

Investment Priorities for AI-Powered Security Platforms

Organizations should prioritize investments in AI solutions that deliver the greatest impact across the seven core AI methods commonly applied in cybersecurity. Machine Learning should be the foundational investment, as it instills discipline and enables systems to scale without human intervention. Neural Networks represent the next priority, helping map relationships between enormous amounts of data and translating input into desired security outputs.

Investment decisions should focus on platforms that integrate multiple AI capabilities, including Expert Systems that emulate human decision-making to solve complex security problems through reasoning and if-then rules. Natural Language Processing (NLP) provides critical capability to understand security-related text with computational linguistics combined with statistical and machine learning models, enabling computers to process human language and automatically perform repetitive security tasks.

Organizations must also invest in Named Entity Recognition (NER) systems that extract information from security data, as contextualization needs to be very specific to understand various security domain terminologies. The substantial computational resources and infrastructure required for AI implementation should be factored into investment planning, as AI algorithms need significant processing power and storage capacity to analyze large volumes of data and perform complex security calculations.

Employee Training for AI-Generated Threat Recognition

With cybercriminals increasingly weaponizing AI technology, employee training must evolve to recognize AI-generated threats and work effectively with AI-powered security systems. Training programs should focus on understanding how AI enhances traditional threat detection methods, moving beyond signature-based detection to pattern recognition that can identify unknown threats.

Employees need to understand that AI-powered systems use machine learning models trained on vast datasets to recognize patterns indicative of malicious behavior. This knowledge helps security professionals interpret AI-generated alerts and understand why certain activities are flagged as suspicious, even when they don’t match known attack signatures.

Training should emphasize the importance of data quality and privacy in AI systems. High-quality data is essential for training accurate AI models to detect and respond to threats, requiring employees to understand data cleansing and validation processes that eliminate errors and inconsistencies that could compromise AI performance. Additionally, staff must be trained on data encryption, anonymization, and access control measures that protect sensitive information while enabling effective threat detection.

Continuous Adaptation Strategies for Evolving Threats

Regular testing and updating of AI models are essential to maintain effectiveness in the dynamic threat landscape. Continuous monitoring of AI performance helps identify areas for improvement and prevents model drift, where AI accuracy degrades over time as new threats emerge.

Organizations should implement scheduled retraining of models with new data to ensure they stay current with emerging threats. This includes conducting adversarial testing to reveal vulnerabilities in AI models, allowing organizations to harden them against potential attacks. Keeping AI models updated and resilient is essential for maintaining adequate cybersecurity defenses.

The adaptation strategy must account for AI’s ability to automate incident response by instantly assessing threat scope and severity, determining appropriate responses, and executing predefined actions such as isolating affected systems or blocking malicious activity. This automation capability requires continuous refinement to ensure responses remain appropriate as attack methods evolve.

Organizations should establish feedback mechanisms that incorporate lessons learned from both successful threat detections and false positives, using this information to continuously improve AI model accuracy and reduce unnecessary alerts that contribute to analyst fatigue.

The AI revolution in cybersecurity has fundamentally transformed both sides of the digital battlefield. While AI empowers defenders with unprecedented threat detection capabilities, processing millions of data points in real-time and identifying subtle anomalies humans would miss, it simultaneously equips cybercriminals with sophisticated tools for creating deepfakes, adaptive malware, and large-scale phishing campaigns. Organizations leveraging AI-driven security platforms report detecting and containing breaches nearly 100 days faster, saving millions in breach costs, yet they must also contend with AI-generated attacks that can bypass traditional defenses.

The future belongs to organizations that embrace AI as both shield and sword while implementing robust governance frameworks to mitigate its risks. Success requires investing in AI-powered defense platforms, maintaining human oversight to combat algorithmic bias and false positives, and continuously adapting strategies as threats evolve. As generative AI democratizes both attack and defense capabilities, standing still means falling behind. The question isn’t whether AI will reshape cybersecurity—it already has—but whether your organization will harness its power to stay ahead of increasingly sophisticated AI-enabled threats.