📊 Full opportunity report: The Frameworks Can’t See the Thing That Matters: A Year of AI-Enabled Cyber Threats on ThorstenMeyerAI.com — validation score, market gap, and execution plan.
TL;DR
A year-long analysis shows AI is increasing cyberattack sophistication and democratizing advanced techniques. Traditional threat assessment methods are now less reliable, raising new security challenges.
A new analysis from Anthropic has found that AI is significantly changing the landscape of cyber threats, making traditional methods of threat assessment less effective. The report examines 832 accounts involved in malicious activity over the past year and shows that AI is enabling less skilled actors to carry out complex, high-risk attacks, which could impact future cybersecurity defenses.
Anthropic’s report analyzed 832 accounts banned for malicious cyber activity between March 2025 and March 2026, mapping their techniques onto the MITRE ATT&CK framework. The key finding is that AI is being used predominantly to automate attack preparation, such as malware creation, with 67.3% of the accounts employing AI for this purpose. More critically, AI is increasingly used for post-breach activities like lateral movement and account discovery, with these techniques rising sharply over the year.
Notably, the report highlights that the use of AI for lateral movement increased from 33% to 56% within six months, indicating a shift toward deeper, more dangerous phases of attacks. AI’s role in activities like account discovery grew by nearly 9%, while its use in phishing declined slightly, suggesting a focus on operational activities once inside a network. This trend indicates that AI is democratizing access to advanced attack techniques, previously limited to highly skilled actors, thereby lowering the technical barriers for malicious actors.
The frameworks can’t see the thing that matters
For decades, danger meant which techniques an attacker commands. A year of real AI-enabled attacks — 832 banned accounts mapped onto MITRE ATT&CK — shows that signal breaking, just as a new, harder-to-see one takes over.
A year of real misuse, mapped to the standard taxonomy
A window, not a census — these are the cases with enough detail to assess techniques thoroughly. Inside it, the risk level climbed fast.
WHAT WAS STUDIED
THE RISK CLIMB · MEDIUM-OR-HIGHER ACTORS
“More techniques” stopped meaning “more dangerous”
The old heuristic: count the techniques, judge the tooling. AI dissolved it — because the model supplies the techniques either way. Watch the old signal fail, then watch what it misses.
Risk score vs. technique count
Two ways to read the same attacker. One is going blind. Press play.
Deeper into the attack — and into less-skilled hands
Across the year, AI use drifted from getting in toward acting once already inside — the operationally demanding stages that used to require an expert.
The attack lifecycle · where AI is now applied
The center of gravity moved right — toward post-compromise work.
From “what they know” to “what they’ve built”
The report sorts the signals into three tiers — one dead, one fading, one durable.
Technique count & tooling
16 vs. 20 between novice and expert; platform doesn’t correlate. The model supplies the techniques either way.
Where in the lifecycle AI is applied
Concentrating on operationally demanding, post-compromise stages is a better signal — but it’s eroding as the whole population heads there.
The scaffolding around the model
Architectures that let the model chain stages and run with minimal human input. Not what they know — whether they’ve built a system that lets AI run the attack.
Fixing the map before the territory moves again
A taxonomy that can’t name the most dangerous behavior on the field will quietly mislead the people relying on it. The response runs in two directions.
Fed back into the models
The findings informed safeguards on the most capable models, built to detect & block some of what was observed:
- Blocking malware development
- Blocking mass data exfiltration
- Putting tools in defenders’ hands first (Project Glasswing)
Taking it to the source
Following the Verizon work, Anthropic says it’s in discussions with MITRE about how ATT&CK might evolve:
- A vocabulary for agentic orchestration
- Naming the scaffolding that makes a model an operator
- An interactive technique visualization on the Red blog
Reading it in proportion
- The 832 cases are a detailed subset, not the full population — the precise percentages are directional, not definitive.
- “More autonomous” is not “fully autonomous” — even the standout case needed human input at key moments, which is itself a place for defenders to intervene.
- This is one vendor’s window — the company with visibility into misuse of its own model, publishing what it found. The right thing to do with the data, and worth remembering as you read it.
How AI Reshapes Threat Assessment and Security Strategies
The findings suggest that traditional threat assessment metrics—such as counting techniques or analyzing tools—are no longer sufficient to gauge attacker danger. As AI enables less skilled actors to perform complex, high-risk activities, the distinction between novice and expert threat actors blurs. This shift could undermine existing cybersecurity defenses and necessitate new approaches focused on behavioral signals and operational patterns rather than technique counts.
AI-POWERED CYBERSECURITY OPERATIONS: Threat intelligence anomaly detection and automated incident response systems
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Evolution of Cyberattack Techniques and AI’s Role
Historically, cybersecurity threat assessment relied on the number of techniques used and the sophistication of tools to estimate attacker risk. Skilled actors employed diverse tactics and specialized tools, making threat levels more predictable. However, recent developments show that AI models like those analyzed by Anthropic are automating complex tasks, reducing the importance of technical skill and tool choice. This change accelerates the threat landscape, aligning with broader trends of AI democratization and automation in cybercrime since early 2025.
“Our analysis shows a marked increase in AI-driven post-breach activities, indicating that attackers are moving deeper into networks more rapidly and with less technical expertise.”
— Anthropic’s research team
OSINT 2.0: Advanced AI Tools and Techniques for Open-Source Intelligence (OSINT 2.0 — Artificial Intelligence for Open-Source Intelligence and Cyber Investigations Book 2)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Unclear Impact of AI on Future Threat Landscape
While the report highlights significant shifts, it remains unclear how these trends will evolve beyond March 2026. The full extent of AI democratization and its impact on threat levels, attacker sophistication, and detection methods is still emerging. It is also uncertain how security defenses will adapt to these changes in the near term.
Applied Network Security Monitoring: Collection, Detection, and Analysis
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Monitoring AI-Driven Attack Trends and Defense Strategies
Cybersecurity researchers and practitioners will need to develop new detection frameworks that focus on attacker behavior and operational signals rather than technique counts. Continued analysis of AI-enabled threats and collaboration between industry and researchers will be essential to anticipate and mitigate future risks. Further studies are expected as AI tools become more accessible and attackers refine their methods.
As an affiliate, we earn on qualifying purchases.
Key Questions
How is AI changing the skills required for cyberattacks?
AI automates many complex tasks, allowing less skilled actors to perform activities that previously required expertise, such as lateral movement and account discovery.
Why are traditional threat assessment methods less effective now?
Because AI enables attackers to perform multiple techniques regardless of their skill level, making technique count and tool analysis less indicative of threat level.
What should cybersecurity teams focus on to detect AI-enabled attacks?
Teams should shift toward behavioral and operational signals, monitoring attack patterns and network activity rather than just techniques or tools used.
Will AI make cyberattacks more frequent or more dangerous?
Both are possible. AI lowers barriers to complex attacks, potentially increasing frequency, and enables deeper, more sophisticated breaches, raising danger levels.
How soon might these trends impact enterprise security?
Signs of these shifts are already evident; organizations should prepare for ongoing evolution in attack techniques and threat assessment methods in the coming months.
Source: ThorstenMeyerAI.com
