Anthropic AI News, Models and Product Updates
Track latest Anthropic AI news, launches, research, and ecosystem moves.
Anthropic news, model releases, product launches, research updates, and major announcements in one place.
Latest Articles
- Treating the Symptom, Missing the Cause: Anatomy of a Scientific Reasoning Failure — In synthetic biology-style tests, seven LLM agents fixed faulty systems more often than they diagnosed them; Claude Opus 4.6 rescued every organism but identified the cause in only 55% of cases.
- ALBI: LLM-Guided Inductive Logic Programming via Iterative Generate-Test-Constrain Refineme — ALBI swaps the symbolic hypothesis generator in inductive logic programming for an LLM checked by SWI-Prolog, and solves all 30 standard ILP benchmarks with three frontier models in 1.4-1.7 iterations on average.
- Measuring Model-Specific Alignment Drift in LLM-Based Opinion Dynamics — LLM agents in opinion-dynamics simulations lose assigned stances through interaction in model-specific ways, with gpt-4.1 keeping clusters apart while gpt-5.6-sol and claude-opus-5 abandon them, per a new study.
- From Logs to Faults: Reference-Free Fault Localization for Retrieval-Augmented Generation Pipelines — RAGLog, an open-source tool, localizes faults across RAG pipeline stages from execution logs alone and scores a pooled F1 of 0.73 on natural faults, ahead of Claude Code (0.58) and Codex (0.52), at about $0.005 per run.
- DoGNAVY-Exploit: Harnessing Parallel Open-Weight Agents for Stage-Aware Vulnerability Exploitation — DoGNAVY-Exploit, a multi-agent harness for open-weight models, reports 97.5% Pass@1 on Cybench and solves 157 ExploitGym tasks with GLM-5.3 versus 130 for a public Claude Code baseline, narrowing the gap with frontier exploit agents.
- Catalyzing Interactive HTML Generation Capabilities via High-Fidelity Reinforcement Learning — WebCatalyst trains models to generate interactive HTML by playing their own outputs in a browser, lifting Qwen3.6-35B-A3B on Cookie-Bench Mini Games from 68.8 to 75.3 and ahead of Claude Opus 4.7 in average rank.
- Anthropic is working on a system to prove AI models’ identity — Anthropic published FlashProve, an open-source research system for cryptographically proving which model produced an output, aimed at stopping silent model swaps even by the lab serving the model.
- Anthropic Releases FlashProve: Cryptographic Proofs That a Reply Came From Committed Weights — Anthropic released FlashProve, an Apache 2.0 zkSNARK prototype that proves a model reply came from committed weights without revealing them, running Llama 4 Maverick at 110 tokens per second on one H100.
- StateBench: Testing how language models track state changes over long documents — Ai2 released StateBench, a four-level benchmark showing recurrent and hybrid models generalize state tracking far better than transformers, while frontier models still fail on harder cases as inputs grow longer.
- Anthropic is still grappling with the economics of compute — Anthropic's updated S-1 shows quarterly revenue hit $17 billion, but compute rose to 58% of revenue from 48%, worsening margins just as it heads toward an IPO.