Science & Tech · Cybersecurity · 1 hr ago
Only 25% of cybersecurity teams use agentic AI for defense
A quarter of surveyed cybersecurity teams actively use agentic AI to help defend their networks.
The findings come from a 2026 survey of security teams by Flashpoint, a threat intelligence company.
Another 20% are running closed pilots, while 35% are considering adoption within the next 12 to 24 months.
These tools can sort information and prioritize alerts, and some can block internet addresses or take other security actions.
Teams worry agents may make confident but wrong decisions, and also cite data leaks, oversight, regulation and cost as concerns.
Many teams want more evidence that the tools can be trusted before giving them greater control over security work.
More than half expect their budgets to rise to support investigation of agentic AI, with 35% expecting an increase of more than 10% over the next year.
Only 25% of surveyed cybersecurity teams actively use agentic AI in their cyber defense strategy, according to a Flashpoint report.
Another 20% are running closed pilot schemes, while 35% are examining adoption within the next 12 to 24 months.
The report says teams have concerns about effectiveness, data leakage, auditability, regulatory requirements and cost.
Some teams use AI agents for tasks including prioritizing alerts and blocking IP addresses, but many are wary of giving them decision-making authority.
More than half of respondents expect their cybersecurity budgets to increase to support investigation of agentic AI.
- Who
- Cybersecurity teams surveyed for Flashpoint’s Enterprise Threat Intelligence Report: 2026.
- What
- The report found that 25% actively use agentic AI for cyber defense, while others are piloting or considering it.
- When
- The report was published on October 13, 2026.
- Where
- Not stated.
- Why
- Teams are weighing the potential of agentic AI against concerns including its effectiveness, data leakage, auditability, regulatory requirements and cost.
This story does not have two clearly opposing sides.
This story does not have a timeline yet.
- Active use
- 25% of surveyed cybersecurity teams
- Closed pilots
- 20% of respondents
- Considering adoption
- 35% are examining adoption within 12 to 24 months
- No plans to test
- 20% of respondents
- Expected budget increase
- 35% expect an increase of over 10% in the next 12 months and 15% expect an increase of up to 10%
What people are saying
An agent that’s confidently wrong is worse than one that’s slow.
The opportunity for agentic AI on the defensive side is significant, and our research shows security leaders are ready to invest. But they aren't swayed by the hype.
They're holding agentic AI to a high bar, and they want to know where it improves their work, whether they can trust the output, and where human judgment must remain part of the process.
Sources
Related news
Cyberattack disrupts refrigeration at hundreds of French supermarkets
Cyberattack disrupts IT systems at three TFG London fashion brands
GitLab fixes maximum-severity flaw that could let attackers run server code
Ledger links CryptoBilis wallet implants to confirmed crypto thefts
Cyberattacks target refrigeration systems at more French supermarkets
Ahold Delhaize takes some systems offline after cyberattack
CISA adds five exploited software flaws amid China-linked espionage concerns
SAP fixes maximum-severity flaw in its Python Cloud SDK
Trellix launches on-premises AI tool for security investigations
Canadian cybersecurity executive arrested in ShinyHunters investigation
Cyberattack disrupts internet services at St. Petersburg provider