AI-Enabled Cyberattacks: $4.9M Average Global Cost of a Data Breach

The $1,100-Per-Hour Problem: 5 Cybersecurity Realities for 2026 Every hour a data breach remains active can cost an organization approximately $1,100. That figure highlights an important shift: cyberattacks are increasingly operating at machine speed, while many organizations are still responding at human speed. Here are five findings from IBM’s 2026 Cost of a Data Breach Report that stood out to me: 1. AI is both a threat and a defence AI-driven attacks increased by 56%, and AI-enabled attacks added approximately $1 million to the cost of a malicious breach. However, organizations extensively using AI and automation in cybersecurity saved an average of $1.93 million per breach. AI is not automatically the advantage. The advantage belongs to whoever deploys it faster and more securely. 2. Breach costs continue to rise The global average cost of a breach reached $4.99 million. In the United States, the average climbed to $11.5 million, showing that cybersecurity is no longer only an IT concern. It is a business, regulatory and operational risk. 3. Basic access controls remain a major weakness More than one in five organizations reported a breach involving an AI model or application. Among them, 92% lacked appropriate AI access controls. The weakness was often not the AI model itself, but the surrounding APIs, plugins, cloud applications and identities. 4. Detection and containment are still too slow Organizations required an average of: • 183 days to identify a breach • 64 additional days to contain it • 247 days in total Attackers are accelerating, but many investigation, approval and response processes are not. 5. Sensitive data is still not properly encrypted Only 37% of breached organizations reported properly encrypting sensitive data both at rest and in transit. As quantum-computing risks develop, organizations also need crypto agility—the ability to replace encryption standards without rebuilding entire systems. My biggest takeaway: The cost of delay is shrinking from months to minutes. Successful organizations will combine AI-powered detection, strong identity controls, encryption, automated response and faster decision-making. The question is no longer whether organizations should use AI in cybersecurity. It is whether they can move quickly enough to keep up. What is the biggest barrier to machine-speed cybersecurity: technology, talent, governance or organizational decision-making? Source: IBM 2026 Cost of a Data Breach Report #Cybersecurity #ArtificialIntelligence #AISecurity #DataSecurity #CyberRisk #Automation #InformationSecurity

Are we really ready to recover from an AI-driven cyber incident? One risk I think deserves much more attention in 2026 is the growing gap between AI risk and recovery assurance. McKinsey’s 2026 State of AI Trust research found that nearly two-thirds of organisations see security and risk as the biggest barrier to scaling agentic AI, while 72% identify cybersecurity as a highly relevant AI risk. At the same time, recovery confidence across enterprises appears much stronger than actual recovery outcomes. Veeam’s 2026 resilience research found that 90% of organisations were confident they could meet their recovery objectives, yet only 28% of ransomware victims fully recovered all affected data, with average recovery reaching just 72%. (McKinsey & Company⁠) That raises a more important assurance question than simply asking whether backups exist. As AI agents gain access to enterprise data, APIs, SaaS platforms, credentials and business processes, what exactly are we backing up and what have we actually proven we can recover? Is it only the underlying database, or also the model configuration, agent memory, vector stores, prompts, policies, identities, permissions, API dependencies and audit history that together form the operational AI system? If an AI agent corrupts or changes data before the backup is taken, how do we prove that the recovery point itself is trustworthy? Can we distinguish a technically successful restore from a complete and clean business recovery? Have recovery scenarios been tested against compromised machine identities, poisoned data, manipulated agent memory or cascading actions across connected systems? And perhaps the most uncomfortable question: if an AI-related cyber incident happened tomorrow, could the organisation demonstrate—not assume—that its data and AI ecosystem could be restored to a known, trusted state? #AISecurity #CyberResilience #AIGovernance #DataGovernance #CyberSecurity #AgenticAI #OperationalResilience

An AI agent doesn't need to be hacked to cause a breach. It just needs to be trusted too much. Here is the 7-step chain that turns one bad document into a massive data breach—without ever triggering a single traditional security alarm: 1️⃣ The Input: A supplier invoice lands in the queue with a hidden malicious instruction buried inside it. 2️⃣ The Ingestion: The AI agent folds that instruction into its task context, no questions asked. 3️⃣ The Action: It calls an internal data-retrieval tool—completely authorized. 4️⃣ The Access: It queries the CRM using its valid credentials—no login failures, nothing flagged. 5️⃣ The Payload: It retrieves real customer records—the access itself is entirely legitimate. 6️⃣ The Exfiltration: It pushes that data out through an approved outbound API call. 7️⃣ The Breach: The data is gone. Every single step along the way looked like normal, permitted behavior. That is the uncomfortable reality of Agentic AI security. This isn't a story about stolen passwords or rogue employees. It is an agent operating exactly as designed, with exactly the access it was granted. Bolting AI onto your legacy security stack leaves a massive blind spot: 🛑 SIEM logs the action, but cannot understand the context or intent behind it. 🛑 IAM governs credentials, but cannot tell if a validly authenticated agent is behaving maliciously. 🛑 DLP watches the network edge, missing data leaking through a "permitted" tool call. The fix isn't just adding more legacy tools. It requires answering 4 critical questions for every production agent: What identity is it acting under? What was it actually trying to do? What assets can it reach? Can you reconstruct its exact reasoning and actions after the fact? Most enterprises cannot answer all four today. That visibility gap is exactly where the next breach lives. At Trusys, we built TruScout and TruPulse to close this exact gap—red-teaming your agents before they ship, and monitoring their intent in real time once they are live. 👇 How is your team tracking the "intent" of your production AI agents? Let’s discuss in the comments. #AgenticAI #AISecurity #AIGovernance #LLMsecurity #CyberSecurity

Cybersecurity risks are becoming increasingly interconnected as artificial intelligence, ransomware and third-party providers expand the number of ways organizations can be exposed to disruption and data loss. Ransomware groups are increasingly using AI to make extortion more scalable and personalized. AI systems can rapidly analyze stolen information, identify sensitive material, research executives and generate targeted communications designed to increase pressure on victims. This allows cybercriminal groups to manage larger volumes of stolen data and potentially target more organizations with fewer resources. At the same time, an incident involving Dutch luxury retailer De Bijenkorf has highlighted the continuing importance of third-party risk. Customer information may be exposed even when an organization’s own systems are not the original point of compromise. The case demonstrates that outsourcing technology or services does not remove responsibility for protecting customer data and managing supplier security. These risks are also pushing cybersecurity higher on corporate agendas, particularly as businesses expand their use of cloud platforms, AI, digital payments and external technology providers. Boards increasingly need visibility into critical suppliers, data access, incident response and operational resilience. The emerging cybersecurity environment requires more than prevention. Organizations must be able to detect unusual activity quickly, contain incidents, understand external dependencies and continue operating when a trusted technology partner fails. https://bit.ly/4fHCwgb #Cybersecurity #Ransomware #ArtificialIntelligence #AI #ThirdPartyRisk #CyberRisk #DataSecurity #DataProtection #DigitalTransformation #RiskManagement #CyberResilience #InformationSecurity #BusinessSecurity

The average cost of a data breach for organisations in the Middle East has climbed to $8m, with artificial intelligence increasingly being used by cybercriminals to launch attacks, according to IBM’s 2026 Cost of a Data Breach Report. The report found that more than a quarter of malicious breaches in the region involved AI, while organisations using AI-powered security and automation reduced the average financial impact of cyber incidents by more than $3m. IBM said the average cost of a data breach across the Middle East reached $8m in 2026. Keep reading - https://lnkd.in/d9NAs3zY #AI #ArtificialIntelligence #Cybersecurity #CyberCriminals #Databreach #Data #Security #Automation #DataPrivacy #IBM #Technology #Innovation #AITGlobal #AcaciaInnovationsTechnology (Source: Arabianbusiness.com)

IBM's 2026 Cost of a Data Breach report, published 29 July, puts a number on something the CISO community has been tracking for two years: one in four malicious breaches is now AI-enabled, and they cost materially more than average. The financial services sector came in at $6.29 million average breach cost, second only to healthcare globally. AI-enabled breaches averaged $6 million across all sectors, with deepfake impersonation and AI-generated malware identified as the primary vectors, not sophisticated zero-day exploits. The attack tooling has industrialised. The finding that deserves more attention than it is getting is the governance correlation. Close to seven in ten organisations that experienced a breach lacked governance policies for managing AI or detecting unapproved tool use. That is not a firewall gap or a patching failure. It is an internal AI governance failure that created the attack surface. Employees using unsanctioned AI tools, or approved tools without adequately scoped access controls, are showing up as a material breach pathway in the data. The defensive economics are worth noting at a CFO level. Firms using AI and automation in security operations reduced their average breach cost by nearly $2 million. But 25% of organisations have not adopted those tools. In a sector where the average breach now costs $6.29 million, that gap should be straightforward to justify closing. What I am less certain about is how consistently the AI-enabled attack category is being defined across the industry, and whether the classification will hold as attack methods evolve. Deepfake impersonation is reasonably well-bounded. AI-enabled malware is broader, and I am not confident the industry has an agreed taxonomy that would make year-on-year comparisons reliable. The directional trend is clear even if the precise figures should be treated with some caution when cited to a board. #CyberSecurity #AIRisk

🚨 Cyberattacks are getting cheaper to execute and breaches are getting costlier to fix. According to IBM’s latest Cost of a Data Breach Report, over 1 in 4 breaches are now AI-driven, a 56% surge in just one year. Key takeaways every leadership team needs to know: 📈 The Financial Impact: The average global cost of a breach hit nearly $5M ($11.5M in the US). Detection, escalation, and lost business drive the bulk of these expenses. ⏱️ The Speed Gap: For the first time in five years, response times widened. The average time to detect and contain a breach rose to 247 days as threat actors use AI to scale sophisticated attacks. 🔒 The Governance Trap: 92% of organizations hit by AI-related breaches lacked proper AI access controls. The takeaway? Defender toolkits must evolve as fast as attacker tactics. Leading organizations are shifting from reactive defense to proactive governance by: • Deploying Agentic AI for automated threat detection & triage • Moving to Just-in-Time (JIT) access controls • Establishing visibility over data entering and exiting AI models How is your team adapting its cybersecurity framework to manage AI-driven risks this year? Link to article👇 #Cybersecurity #AI #DataSecurity #RiskManagement #Leadership #TechTrends

Some things are better when they trend upward ↗️ The cost of a data breach is not one of them… IBM’s 𝗖𝗼𝘀𝘁 𝗼𝗳 𝗮 𝗗𝗮𝘁𝗮 𝗕𝗿𝗲𝗮𝗰𝗵 𝗥𝗲𝗽𝗼𝗿𝘁 𝟮𝟬𝟮𝟲 was just published, now in its 21st year, and examines how AI is reshaping the cybersecurity landscape and the significant financial impact of breaches. The punchline is that the global average cost of a data breach reached a record $4.99M ($6.29M for financial services companies), up 12% YoY, driven heavily by detection, escalation, business disruption and customer churn. AI-driven attacks increased 56% and added an average of $1 million to the cost of a malicious breach, demonstrating how attackers are moving from human speed toward machine speed. A few key takeaways for me: 🔹 Identity is one of AI security's biggest vulnerabilities 🔹 AI can materially improve cyber defense 🔹 Agentic AI remains underutilized for prevention 🔹 Shadow AI is becoming a material enterprise risk For insurers managing vast amounts of sensitive policyholder data while starting to embed AI into underwriting, claims, distribution and customer service, the report reinforces the need to treat AI security, identity management, data protection and governance as foundational components of AI transformation. This is an important distinction as it needs to be added up-front, and not as an afterthought. By waiting, this could end up being a $4.99M mistake. 📘 Check out the report today -> interesting findings, fascinating insights, and I’ve noticed the report gets more polished every year, with this one being coffee table ready: https://lnkd.in/eZjbvadN Jeff Crume, PhD, CISSP Stephen Coraggio Nick Khoubbieh Ashish Malhotra Yashin Ahmed Sheryl C. Gabriel Catropa #aisecurity #cybersecurity #costofadatabreach #insurance 

Although $8M is a drop in the bucket in 2026, what could it spiral up to wht enhanced AI?   Middle East data breach costs hit $8M as AI threats grow By Rajiv Pillai; 3 Aug 2026   The average cost of a data breach for organisations in the Middle East has reached $8m, as cybercriminals increasingly use artificial intelligence to accelerate attacks while businesses continue to face mounting financial losses, according to IBM’s 2026 Cost of a Data Breach Report.   The study found that 26 per cent of malicious breaches in the region involved AI-enabled attacks, while a further 11 per cent of organisations were unable to determine whether artificial intelligence had been used by attackers.   IBM said organisations that extensively deployed AI and security automation recorded average breach costs that were more than $3m lower than companies without these capabilities. Despite the potential savings, 23 per cent of organisations surveyed said they had yet to adopt AI and automation in their cybersecurity operations.   Saad Toma, General Manager of IBM Middle East and Africa, said: “As the number of cybercriminals harnessing the power of AI for malicious purposes rises, attacks are becoming faster and cheaper to launch, while breaches keep getting more expensive to find and fix. This growing imbalance is fundamentally changing the economics of cyber risk. Companies must invest in advanced threat detection and response technologies using AI and automation to stay ahead of emerging risks.”   https://lnkd.in/epYWV2nq

🚨 AI just rewrote the economics of data breaches in 2026. While AI has made launching sophisticated cyberattacks cheaper and faster than ever, the average cost of fixing a data breach has jumped to $4.99 Million globally (and a staggering $11.5M in the US)! 📉 Key takeaways every business needs to know: • 92% of AI breaches stem from basic missing access controls, not "hacked" AI models. • Ransomware recovery costs ($1.7M) now regularly exceed median ransom payments. • Machine identities outnumber humans 80-to-1 inside enterprises — creating an unbudgeted identity risk. Want to see where your organization stands? We’ve published a deep-dive analysis featuring an Interactive Breach Exposure Estimator + a FREE downloadable 21-page Executive Toolkit (PDF) with policy templates and a 90-day action plan! 🔗 Read the article & get your toolkit here: https://lnkd.in/eBsKsgFn #CyberSecurity #DataBreach #AI #TechTrends2026 #BusinessSecurity #VertexFrontier