AI Enabled Cyber Threats Mitre Attack

Counting an attacker's techniques can miss how effectively AI connects those techniques into an operation. Anthropic examined 832 accounts banned for malicious cyber activity between March 2025 and March 2026, selecting cases with enough detail for assessment. Its analysis summary says the least-skilled actors used about sixteen techniques on average, compared with about twenty for the most skilled.

The company identifies a more useful distinction in the software surrounding the model: higher-risk actors built systems that chained attack stages together and reduced the need for human decisions. Its conclusion is that assessments should examine this automation, alongside the techniques used. The findings describe the selected banned accounts.