📊 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 empowering cyberattackers to become more dangerous and harder to identify using traditional threat assessment methods. Attackers now perform complex activities with less skill, challenging existing security frameworks.
A new analysis from Anthropic shows AI is significantly changing the landscape of cyber threats, with attackers increasingly using AI to perform complex tasks that previously required high technical skill. This development challenges longstanding threat assessment frameworks and raises concerns about the ability of security teams to accurately identify and respond to danger.
Anthropic examined 832 accounts banned for malicious activity between March 2025 and March 2026, mapping their techniques onto the MITRE ATT&CK framework. The findings reveal that AI is primarily used to automate attack preparation, such as malware creation, with 67.3% of actors employing AI for this purpose. More notably, a growing share of attackers are leveraging AI for sophisticated, post-breach activities like lateral movement within networks, which increased from 33% to 56% over the year.
Furthermore, the report indicates that AI use has shifted from initial access tactics, such as phishing, to deeper network exploitation. AI-driven account discovery and lateral movement techniques have risen, making attacks more dangerous and accessible to less skilled actors. This trend signifies a democratization of offensive capabilities, eroding the traditional link between attacker skill level and threat severity.
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
INTELLIGENT CYBERSECURITY SOFTWARE SYSTEMS: Threat detection automated response and adaptive defense architectures
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.
Network Intrusion Detection
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.
The 72-Hour Family Survival Guide (US-Iran War Edition): The Clock is Ticking. Act Now Before Phones Fail, Stores Empty, and Help Never Comes
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.
Why AI-Enhanced Attacks Undermine Threat Assessment
This shift matters because it challenges the core assumption that threat level correlates with the number of techniques used or the sophistication of tools. As AI automates complex tasks, even less skilled actors can execute high-impact attacks, complicating detection and response. Security frameworks relying on skill-based heuristics may no longer suffice, increasing the risk of undetected breaches and expanding the threat landscape.
Evolution of Cyberattack Techniques with AI Integration
Historically, threat assessment focused on counting techniques and analyzing tool sophistication, assuming that more techniques indicated higher danger. However, recent developments show attackers are increasingly using AI to perform complex tasks, such as lateral movement and privilege escalation, which previously required expertise. The rise of AI-enabled attack automation began in 2024, with a noticeable acceleration in 2025, reflecting broader adoption of AI models in cybercrime.
The report from Anthropic builds on prior concerns about AI’s dual-use nature, highlighting how malicious actors are leveraging these tools to bypass traditional defenses and democratize offensive capabilities across different skill levels.
“Attackers are shifting their focus from simple entry techniques to complex, post-compromise activities, making threats more dangerous and accessible to a broader range of actors.”
— Anthropic’s research team
Limitations and Unanswered Questions About AI-Driven Threats
While the report provides a significant window into current trends, it notes that the data is based on a subset of banned accounts where detailed technique mapping was possible. It remains unclear how representative this sample is of the entire threat landscape. Additionally, the long-term impact of AI on threat detection and the development of countermeasures are still evolving areas, with ongoing uncertainty about how quickly defenses can adapt.
Future Steps for Security Teams and Policy Makers
Security practitioners will need to update threat assessment frameworks to account for AI-enabled attack techniques and focus on behavioral signals rather than technique counts alone. Developing AI-specific detection tools and investing in threat intelligence that monitors AI-driven attack patterns will be critical. Policymakers may also consider regulations to limit malicious AI use and promote responsible development of offensive AI capabilities.
Key Questions
How is AI changing the skills required for cyberattackers?
AI automates complex attack activities, reducing the need for high technical skill and allowing less experienced actors to perform sophisticated operations.
Why are traditional threat assessment methods becoming less effective?
Because AI enables even less skilled attackers to carry out complex techniques, making the number of techniques or tools used less indicative of threat level.
What can organizations do to defend against AI-enabled attacks?
Organizations should update detection systems to focus on behavioral patterns and AI-driven attack indicators, and consider investing in AI-aware cybersecurity tools.
Will AI make cyber threats more widespread?
Yes, as AI lowers the technical barriers to executing advanced attacks, more actors can participate, increasing the overall threat volume.
Source: ThorstenMeyerAI.com
