📊 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 transforming cyberattack techniques, making even less skilled actors more dangerous. Traditional threat metrics no longer reliably differentiate threat levels, raising new security concerns.
Recent research from Anthropic indicates that AI is fundamentally changing cyber threat dynamics, making attackers more capable and rendering traditional threat assessment metrics obsolete. The report, based on an analysis of over 800 banned accounts involved in malicious activity, finds that AI is increasingly used for complex tasks inside compromised networks, even by less skilled actors. This shift challenges long-standing assumptions about what makes a threat dangerous and how security teams should evaluate risk.
Anthropic examined 832 accounts banned for malicious cyber activity between March 2025 and March 2026, mapping their techniques onto the MITRE ATT&CK framework. The findings show that 67.3% of these actors used AI primarily to prepare for attacks, such as malware creation. Notably, AI’s role in post-breach activities like lateral movement increased significantly over the year, with a 70% rise in high-risk behaviors. The trend indicates that AI enables less skilled actors to perform technically demanding operations previously limited to experts, blurring the lines between novice and advanced attackers. Furthermore, traditional indicators of threat level—such as the number of techniques used or the platform employed—no longer reliably predict danger, as even low-skill actors now mimic high-skill behaviors through AI assistance. The report emphasizes that the core signals security teams rely on are now less effective, requiring a reevaluation of threat assessment strategies.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
Python Scripting for Cybersecurity: Linux Edition: Volume 2 – Log Analysis, Network Visibility, and Threat Detection with Hands-On Python Projects
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
“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.
OSINT 2.0: AI-Powered Open-Source Intelligence for Beginners (OSINT 2.0 — Artificial Intelligence for Open-Source Intelligence and Cyber Investigations Book 1)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
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.
The Practice of Network Security Monitoring: Understanding Incident Detection and Response
Used Book in Good Condition
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
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.
cyber threat risk assessment tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
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.
Implications of AI’s Role in Elevating Threat Levels
This development signifies a major shift in cybersecurity, as AI democratizes advanced attack techniques, making it easier for less skilled actors to carry out sophisticated operations. Traditional metrics, like technique diversity and tooling, no longer distinguish between high- and low-risk actors. This undermines existing threat models and calls for new detection strategies. The findings highlight an urgent need for security professionals to adapt to AI-driven threats, as the landscape becomes more unpredictable and dangerous. Failure to do so could result in increased successful breaches and broader cyber risks for organizations worldwide.Evolution of Cyberattack Techniques and AI Integration
For decades, threat assessment relied on the assumption that more techniques and advanced tools indicated higher danger. The MITRE ATT&CK framework provided a structured way to categorize attacker behaviors. However, recent developments show that AI has begun to automate complex tasks, enabling even less skilled actors to mimic highly capable attackers. The trend accelerated over the past year, with attackers shifting focus from initial access techniques to deeper network operations. This evolution reflects broader AI adoption in cybercrime, transforming the threat landscape in ways that current frameworks do not adequately capture.“AI is effectively lowering the skill barrier, allowing less experienced actors to perform operations that once required significant expertise.”
— Thorsten Meyer, researcher at Anthropic
Unclear Impact of Evolving AI-Driven Threats
It remains unclear how quickly security frameworks can adapt to these changes or what new metrics will effectively differentiate threat levels. The full scope of AI’s use in cybercrime is still emerging, and the long-term implications for threat assessment are uncertain. Additionally, the extent to which defenders can counteract AI-enabled attacks with new tools is not yet known.Next Steps for Cybersecurity in an AI-Driven Era
Security organizations will need to develop new detection and assessment methods that account for AI-enabled attack behaviors. Researchers are likely to focus on identifying durable signals that distinguish genuine threat levels despite AI assistance. Additionally, policymakers and industry leaders may consider updating standards and frameworks to better reflect the evolving threat landscape. Continued monitoring of AI’s role in cybercrime will be essential to stay ahead of attackers.Key Questions
How is AI making cyber attackers more dangerous?
AI automates complex tasks like lateral movement and account discovery, which previously required skilled hackers. This allows less experienced actors to perform sophisticated operations inside networks.
Why are traditional threat assessment methods no longer effective?
Because AI enables even low-skill actors to mimic high-skill behaviors, metrics like the number of techniques used or tool choice no longer reliably indicate threat level.
What can organizations do to improve detection of AI-enabled threats?
Organizations need to develop new indicators that focus on operational behaviors and the context of attack activities, rather than relying solely on technique counts or tooling used.
How quickly might security frameworks adapt to these changes?
The timeline is uncertain. Developing new assessment methods and updating existing frameworks will require concerted effort, research, and industry collaboration.
Source: ThorstenMeyerAI.com
