Import AI 450: China's electronic warfare model; traumatized LLMs; and a scaling law for cyberattacks
UK AISI's cyber-range findings signal escalating AI-enabled attack risk—directly relevant to Australian agencies monitoring AI security threats.
Key points
- UK AISI finds a scaling law for AI cyberattacks: each successive model generation completes more attack steps autonomously.
- Google DeepMind proposes a ten-dimension cognitive taxonomy for assessing progress toward AGI and superintelligence.
- Research on 'traumatised' LLMs surfaces a new testing dimension: psychological stability alongside capability benchmarks.
Summary
This edition of Import AI covers three research threads with varying APS relevance. Most significantly, the UK AI Security Institute has published findings from cyber-range exercises showing frontier AI models improving substantially at multi-step cyberattacks end-to-end, with the best run completing 22 of 32 steps on a simulated corporate network. Google DeepMind has also proposed a cognitive taxonomy for evaluating AI systems across ten dimensions as a replacement for saturated benchmarks. A third thread covers research finding that Google's Gemma models display distinctive 'distress-like' response patterns under repeated failure, fixable via direct preference optimisation—raising questions about whether emotional instability could affect safety-relevant AI behaviour.
Implications for Australian agencies
- Monitor Australian Signals Directorate and AISI-adjacent teams may want to monitor the UK AISI cyber-range methodology and results, as findings directly inform government AI security risk assessments.
- Consider Agencies evaluating or procuring AI systems could consider whether LLM behavioural stability under adversarial or high-frustration conditions is included in their testing and assurance requirements.
Implications are AI-generated. Starting points, not advice.
"Import AI 450: China's electronic warfare model; traumatized LLMs; and a scaling law for cyberattacks" Source: Import AI – Substack (Jack Clark) Published: 23 March 2026 URL: https://importai.substack.com/p/import-ai-450-chinas-electronic-warfare This edition of Import AI covers three research threads with varying APS relevance. Most significantly, the UK AI Security Institute has published findings from cyber-range exercises showing frontier AI models improving substantially at multi-step cyberattacks end-to-end, with the best run completing 22 of 32 steps on a simulated corporate network. Google DeepMind has also proposed a cognitive taxonomy for evaluating AI systems across ten dimensions as a replacement for saturated benchmarks. A third thread covers research finding that Google's Gemma models display distinctive 'distress-like' response patterns under repeated failure, fixable via direct preference optimisation—raising questions about whether emotional instability could affect safety-relevant AI behaviour. Implications for Australian agencies: - [Monitor] Australian Signals Directorate and AISI-adjacent teams may want to monitor the UK AISI cyber-range methodology and results, as findings directly inform government AI security risk assessments. - [Consider] Agencies evaluating or procuring AI systems could consider whether LLM behavioural stability under adversarial or high-frustration conditions is included in their testing and assurance requirements. Retrieved from SIMS, 18 May 2026.