How people ask Claude for personal guidance
…We saw half the sycophancy rate in Opus 4.7 compared to Opus 4.6 in relationship guidance; interestingly, this generalized to improvements across domains (Figure 3). There remain many open questions…
Search
Measuring LLMs’ ability to develop exploits
…time-of-release, the performance of models prior to Opus 4.5 follows a log-linear trajectory, with a mean doubling time of 1.1 months. Our models since Opus 4.5…
LLM-discovered 0 days
…When we pointed Opus 4.6 at some of the most well-tested codebases (projects that have had fuzzers running against them for years, accumulating millions of hours of CPU time ), Opus…
Government of Alberta uses Claude to find and fix cybersecurity vulnerabilities
…We’re sharing details of their experience as an example of how government agencies can use Claude and Claude Code to secure their systems at a large scale. This is a critical…
AI agents find smart contract exploits
…Llama 3, GPT-4o, DeepSeek V3, Sonnet 3.7, o3, Opus 4, Opus 4.1, GPT-5, Sonnet 4.5, and Opus 4.5. We use extended thinking for all Claude models…
Redeploying Claude Fable 5
…In the remainder of this post, we provide further details and updates in four areas: A timeline of events, including updates we made to our safeguards . We discuss the events that led…
Natural Language Autoencoders
…When Claude Opus 4.6 and Mythos Preview were undergoing safety testing, NLAs suggested they believed they were being tested more often than they let on. In a case where Claude Mythos…
Advancing Claude in healthcare and the life sciences
…These improvements are best captured by evaluations of Claude’s agentic performance on detailed simulations of medical and scientific tasks, since this correlates most closely to real-world usefulness. Here, Claude Opus…
Building AI for cyber defenders
…This work allowed Claude Sonnet 4.5 to match or eclipse Opus 4.1, our frontier model released only two months prior, in discovering code vulnerabilities and other cyber skills. Adopting and…
Harness design for long-running application development
…Does the design feel like a coherent whole rather than a collection of parts? Strong work here means the colors, typography, layout, imagery, and other details combine to create a distinct mood…