Insider Threat Statistics 2026: Costs, Causes and AI Risk
Insider risk now costs the average large organization $19.5 million a year, according to the 2026 Ponemon Institute study sponsored by DTEX, up from $17.4 million a year earlier and $16.2 million in 2023. More than half of insider incidents are caused by negligence rather than malice, the average incident takes 67 days to contain, and the fastest-growing source of exposure is employees putting company data into AI tools through personal accounts. Verizon's 2026 Data Breach Investigations Report found 62% of breaches involved a human element and 12% an internal actor; IBM's 2026 Cost of a Data Breach report found shadow AI involved in 43% of breaches.
This page collects the most current figures from the primary reports, with the date, sample and definition behind each, so the numbers can be cited accurately. Vendor-sponsored surveys are labeled as such. For how employers monitor for insider risk without overreaching, see user activity monitoring and the ethical monitoring playbook.
Headline figures
- $19.5 million: average annual cost of insider risk per organization in 2025 (Ponemon/DTEX, February 24, 2026; 354 organizations).
- 25.4: average insider incidents per organization per year (same study).
- About 54%: share of insider incidents caused by negligence, 13.8 per organization a year (same study).
- 67 days: average time to contain an insider incident, down from 81 in the 2025 study and 86 in 2023.
- 62%: share of breaches involving a human element in Verizon's 2026 DBIR; 12% involved an internal actor.
- 90%: organizations reporting at least one insider incident in the past year (Gurucul and Cybersecurity Insiders, March 2026; 725 respondents).
- 43%: share of breaches in IBM's 2026 study involving shadow AI, up from 20%; average cost $5.39 million.
- 39.7%: share of employee interactions with AI tools that involved sensitive data (Cyberhaven, 2026).
What insider incidents cost
The Ponemon Institute's 2026 Cost of Insider Risks study, sponsored by DTEX Systems and based on interviews with 8,750 practitioners at 354 organizations with fieldwork through September 2025, is the most widely cited cost benchmark. Its findings:
- Average annual cost of insider risk: $19.5 million in 2025, up from $17.4 million (2024) and $16.2 million (2023), a 20% rise over two years.
- Negligent incidents cost about $747,107 each and $10.3 million a year per organization, up 17% year on year.
- Containment is the largest per-incident cost activity at $247,587; escalation costs about $39,728.
- Organizations that contain incidents within 30 days spend $14.2 million a year on average; those taking more than 90 days spend $21.9 million.
- In the 2025 edition (8,306 practitioners, 349 organizations), health care and pharmaceuticals had the highest activity cost at $29.2 million, followed by technology at $23 million.
IBM's 2026 Cost of a Data Breach report (July 29, 2026; 602 organizations) puts the global average breach at a record $4.99 million, up 12%, and the US average at $11.5 million; health care breaches averaged $6.64 million and financial services $6.29 million. The 2025 edition found malicious-insider breaches among the most expensive, at $4.92 million; the 2026 figure for that category was not in the published summaries. Mimecast's State of Human Risk 2026 (2,500 respondents, nine countries) estimated that respondents' organizations saw about six insider-driven incidents a month at an estimated $13.1 million each, a self-reported figure much higher than the Ponemon per-incident costs and best read as perceived exposure.
Negligent, malicious and compromised insiders
- The 2025 Ponemon study split incidents 55% negligent or mistaken, 25% malicious or criminal and 20% "outsmarted" insiders whose credentials were stolen. The 2026 edition reports negligence at 13.8 of 25.4 incidents, about 54%.
- In the Gurucul 2026 Insider Risk Report, 74% of security leaders ranked negligent insiders as a top concern, against 59% for malicious insiders; the share of organizations reporting zero incidents fell from 17% to 10%, and the share reporting more than 20 a year doubled.
- Mimecast found 42% of organizations saw malicious insider incidents rise in the past year (33% in 2024) and 66% expected insider-driven data loss to rise in the next 12 months.
- Exabeam's August 2025 survey found 64% of security professionals saw insiders as a bigger risk than external attackers.
- Verizon's 2026 DBIR found insiders' motives were convenience (60%) and financial gain (33%), which supports the picture of most insider risk as shortcuts rather than sabotage.
Detection and containment time
- Average days to contain an insider incident: 86 (2023), 81 (2025 study), 67 (2026 study) (Ponemon/DTEX).
- IBM 2026: mean time to identify and contain a breach of any kind rose to 247 days (183 to identify, 64 to contain), the first increase in five years.
- Detection difficulty: in Securonix-sponsored research, about 90% of respondents said insider attacks were as hard or harder to detect than external attacks (secondary citation; primary not checked).
The cost gap between incidents contained in 30 days and those taking 90 or more, $7.7 million a year in the Ponemon data, is the main economic argument for investing in detection and response rather than prevention alone.
What the 2026 Verizon DBIR found
Verizon's 2026 Data Breach Investigations Report analyzed more than 31,000 incidents and 22,000 confirmed breaches from November 2024 to October 2025.
- Human element in 62% of breaches (60% in 2025, 68% in 2024, 74% in 2023; definitions changed between editions, so the trend is approximate).
- Internal actors in 12% of breaches, down from 18%.
- Breach patterns: system intrusion 61%, social engineering 17%, basic web application attacks 10%, miscellaneous errors 8% (12% in 2025), privilege misuse 3% (7% in 2025).
- Third-party involvement in 48% of breaches; ransomware in 48%; exploitation of vulnerabilities was the top initial access vector at 31%, ahead of stolen credentials.
- 67% of users accessing AI services on corporate devices did so with non-corporate accounts; 45% were regular AI users on corporate devices, up from 15%; more than 15% of corporate users had unauthorized AI browser extensions. Shadow AI was the third most common non-malicious insider action caught by data loss prevention tools.
Shadow AI and data exposure
- IBM 2026: shadow AI involved in 43% of breaches (20% in 2025), with an average cost of $5.39 million; AI-related breaches overall 21%; one in four malicious breaches was AI-enabled, averaging $6.04 million.
- Ponemon/DTEX 2026: 92% said generative AI had changed how employees access and share data; 73% worried unauthorized AI creates invisible exfiltration paths; only 18% had integrated AI governance into their insider risk programs.
- Cyberhaven 2026 (222 companies): 39.7% of AI interactions involved sensitive data; employees entered proprietary data into AI tools about once every three days; a third used personal accounts.
- Cyberhaven 2025: 34.8% of corporate data put into AI tools was sensitive, up from 27.4% in 2024 and 10.7% in 2023; source code was 18.7% of it.
- Exabeam 2025: 76% of security teams reported shadow AI in use.
For what employers can and cannot see of AI use, see can my employer see if I use ChatGPT and the AI use policy template.
By industry
Share of breaches involving internal actors in the 2026 DBIR, by sector:
| Sector | Internal actor share | Human element | Note |
|---|---|---|---|
| Public administration | 44% | n/a | Miscellaneous errors and privilege misuse in top three patterns |
| Education | 22% | n/a | |
| Health care | 19% | 54% | Misdelivery and misuse remain common |
| Financial and insurance | 12% | 65% | |
| Manufacturing | 5% | n/a | Dominated by external system intrusion |
| Retail | 1% | n/a |
Departing employees
- A 2026 Sagiss survey found 64% of former employees said they retained access to sensitive company data after leaving.
- Code42 research (older, vendor) estimated a 37% chance of losing intellectual property when an employee quits and found 71% of firms did not know what data leavers took.
- Departures cluster risk: the weeks between resignation and last day are when most data-exfiltration cases are detected, which is why offboarding checklists revoke access on the last day, not after. The offboarding checklist has the steps.
Insider risk programs and budgets
- 63% of organizations run a dedicated insider risk program; programs prevented about seven incidents and avoided $8.2 million on average (Ponemon/DTEX 2026).
- Insider risk spending rose from 8.2% of the IT security budget in 2023 to about 19% in 2025; 64% increased budgets and 45% still said funding was insufficient.
- Exabeam 2025: 88% had insider programs, but only 44% used user and entity behavior analytics; 74% said executives underestimate insider risk.
Definitions and how to read the numbers
The Cybersecurity and Infrastructure Security Agency defines an insider as "any person who has or had authorized access to or knowledge of an organization's resources" and an insider threat as "the potential for an insider to use their authorized access or understanding of an organization to harm that organization," whether unintentionally (negligence, accident), intentionally, in collusion or through a third party. Three cautions when citing the figures:
- Different denominators. Ponemon counts incidents and costs per organization in a sample of large firms; Verizon counts confirmed breaches; IBM costs breaches of all causes. A "12% of breaches" figure and a "54% of incidents" figure are not in conflict.
- Sponsorship. Ponemon/DTEX, Gurucul, Mimecast, Exabeam, Cyberhaven and Code42 sell insider risk products; their surveys are useful for direction and scale but are not independent.
- Self-report. Survey questions about incident counts and costs measure what respondents believe; Verizon's and IBM's are built from case data and are firmer.
Key takeaways
- Insider risk cost the average large organization $19.5 million in 2025, up from $17.4 million and $16.2 million in the two prior years (Ponemon/DTEX, February 2026).
- Most insider incidents are negligent, not malicious: about 54% negligent, 25% malicious and 20% credential theft; Verizon finds insider motives are mostly convenience (60%) and money (33%).
- Average containment time fell to 67 days; organizations containing incidents within 30 days spend $14.2 million a year against $21.9 million for those taking over 90.
- Verizon's 2026 DBIR: 62% of breaches involve the human element, 12% an internal actor, 3% privilege misuse; 67% of employees using AI on corporate devices do so through personal accounts.
- Shadow AI is the fastest-growing exposure: IBM found it in 43% of 2026 breaches (20% in 2025) at $5.39 million each; Cyberhaven found 39.7% of AI interactions involve sensitive data.
- Public administration (44%), education (22%) and health care (19%) have the highest internal-actor shares of breaches; retail and manufacturing the lowest.
- Most figures come from vendor-sponsored surveys; cite the source and sample, and do not compare per-incident and per-breach numbers directly.
Frequently asked questions
How much do insider threats cost?
The Ponemon Institute's 2026 study put the average annual cost of insider risk at $19.5 million per organization in 2025, across an average of 25.4 incidents. Negligent incidents averaged about $747,000 each. IBM's 2026 report put the global average cost of any data breach at $4.99 million and the US average at $11.5 million.
What percentage of data breaches are caused by insiders?
Verizon's 2026 Data Breach Investigations Report found internal actors in 12% of breaches, down from 18% the year before, and a human element (errors, misuse, social engineering) in 62%. Insider share varies by sector, from 44% in public administration to 1% in retail.
Are most insider threats malicious?
No. In Ponemon's data about 54 to 55% of insider incidents are negligent or mistaken, about 25% malicious and about 20% caused by stolen credentials. Verizon found insiders' motives were convenience in 60% of cases and financial gain in 33%.
How long does it take to contain an insider incident?
67 days on average in the 2026 Ponemon study, down from 81 days in 2025 and 86 in 2023. Incidents contained within 30 days cost organizations far less ($14.2 million a year) than those taking over 90 days ($21.9 million).
How does AI affect insider risk?
It creates new exfiltration paths. IBM found shadow AI involved in 43% of 2026 breaches, Verizon found 67% of employees using AI on corporate devices used personal accounts, and Cyberhaven found 39.7% of AI interactions involved sensitive data. Only 18% of organizations had integrated AI governance into insider risk programs.
What is an insider threat?
CISA defines it as the potential for a current or former employee, contractor or partner with authorized access or knowledge to use it to harm the organization, whether through negligence, accident, intentional misconduct, collusion or a third party's use of their credentials.
Which industries have the most insider breaches?
In the 2026 Verizon DBIR, public administration had the highest internal-actor share at 44% of breaches, followed by education at 22% and health care at 19%. Financial services was 12%, manufacturing 5% and retail 1%. Ponemon found health care and pharmaceuticals had the highest insider-risk costs.
Do insider risk programs work?
The Ponemon 2026 data says organizations with dedicated programs (63% of the sample) prevented about seven incidents and avoided about $8.2 million a year on average. Insider-risk spending rose from 8.2% to about 19% of security budgets between 2023 and 2025.