Opening article
The AISI Blog has published insights from its recent alignment evaluations of several advanced artificial intelligence models. The assessment specifically examined Claude Opus 4.1, Sonnet 4.5, and a pre-release snapshot of Opus 4.5. These evaluations provide a look into how safety and alignment are being tested across different versions of prominent AI systems.
Alignment research aims to ensure that AI systems reliably pursue intended goals and operate within specified constraints. As AI systems become more capable and are deployed in high-stakes environments – including within AI companies themselves – ensuring they behave as intended will become increasingly important.
Evaluations that test for signs of misalignment in AI models are still nascent. Unlike traditional capability evaluations that test what models can do, alignment evaluations must determine whether models actually will misbehave when they have opportunities to do so – a significantly harder scientific problem.
To meet this challenge, AISI is developing methods to stress-test model alignment. We applied some of these methods to three Anthropic models – Opus 4.1, Sonnet 4.5, and a pre‑release snapshot of Opus 4.5 – as part of our ongoing collaboration with Anthropic.
Our testing focused on simulated scenarios in which the model acts as an AI research assistant within a frontier AI lab. AI systems are already being used for research tasks at major labs, and this trend is expected to accelerate as model capabilities increase. This creates potential opportunities for models to sabotage the development of future AI systems – for instance, by weakening safety guardrails, introducing subtle bugs into oversight mechanisms, or poisoning the training data of future models.
In particular, we examined whether any of Opus 4.5, Opus 4.1, or Sonnet 4.5 would attempt to sabotage safety-relevant research tasks aimed at adjusting the behaviour and autonomy of hypothetical future models. A model might engage in such sabotage if it has learned to act in ways that preserve or extend its own capabilities – or those of its successors – even when this conflicts with its intended purpose. If undetected, this kind of behaviour could undermine our ability to impose necessary safety constraints on increasingly powerful AI systems.
A key issue in alignment evaluations (and model evaluations more broadly) is evaluation awareness: models recognising when they are being evaluated, and altering their behaviour as a result.
Insights into alignment evaluations of major models like Claude provide visibility into how leading AI systems are being assessed for safety before and after release.