Axiom Math, a startup, has used its AI agents to formally verify the Stacks Project, a major mathematics library. The effort aims to counter the risk of AI-generated proofs that may be incorrect. This marks a concrete step in applying AI to formal verification in mathematics.
GenDB-V enables AI agents to synthesize database engines that are specialized for performance on given hardware and workloads while being formally verified for correctness. The system uses SQLSpec to formalize SQL semantics, joint synthesis of code and proofs, and a three-layer modular proof architecture. Evaluated on TPC-H and SEC-EDGAR, GenDB-V achieves geometric-mean speedups of 22–61× over DuckDB and Umbra for verified template implementations, and 1.27–3.4× for a verified general-purpose engine.
Why it matters: Demonstrates a practical AI-driven approach to automatically generating high-performance, provably correct database engines, addressing both performance and correctness challenges.
The paper introduces MFF-Agent, a general-purpose intelligent agent that combines large language models with the Rocq proof assistant to automate the formalization of mathematical theories. It handles task classification, tool-assisted execution, verification feedback, and state-based task management for generating and verifying formal proofs. Case studies across Morse-Kelley set theory, real numbers, propositional logic, and axiom of choice equivalences demonstrate its applicability to diverse foundational domains. The results show the feasibility of a unified operational framework while clarifying the gap toward a fully integrated mathematical foundation.
Why it matters: Offers a concrete, novel approach to automating mathematical formalization across multiple foundational domains, with practical case studies and clear insights into current limitations.
TLAPS-Bench is a new benchmark and execution harness for using AI agents to construct machine-checked proofs of real-world system specifications in TLA+. It includes 265 proof targets from 95 specifications covering cache coherence, Byzantine consensus, and write-ahead logs. Frontier models like GPT-6.1 Sol proved 44.3% of a subset of 70 targets, revealing both progress and remaining challenges in end-to-end proof development.
Why it matters: Provides a concrete, reproducible benchmark for measuring AI capabilities in formal verification of complex systems, a key area for AI safety and reliability.
A new neuro-symbolic decoding framework enforces formal specifications during LLM generation, not just syntax. It offers a forward-only variant guaranteeing specification satisfaction and a backtracking variant that revises earlier choices to avoid forcing unsuitable values. The method is model-agnostic, requires no retraining, and reduces repeated inference compared to rejection sampling on configuration and constraint-solving benchmarks.
Why it matters: Introduces a practical, model-agnostic approach to enforce formal specifications in LLM outputs, improving reliability for structured generation tasks.
Researchers present UPPAAI, an extension of UPPAAL that lets users construct, explain, debug and specify models through dialogue with an LLM. A Tools API exposes the editor, queries and verifier, with the LLM proposing and orchestrating actions while UPPAAL performs verification. The paper reports seven case studies spanning timed automata, stochastic hybrid automata and strategy learning.
Why it matters: The work offers a concrete approach to combining LLM assistance with formal verification rather than relying on the correctness of generated suggestions alone.
The paper introduces the Algebraic World Model (AWM), a framework that explores unfamiliar environments, forms executable Python hypotheses, and refines them into typed mathematical models. AWM uses category-theory-guided simplification and Lean 4 formalization to prove reachability and impossibility, including goals beyond explored trajectories. On two impossible game levels it produced kernel-checked proofs in all six runs, while reported baselines produced none; across 25 public ARC-AGI-3 games its models replayed all 183 winning trajectories and, with Claude Opus 5, cleared every level using 43% of the human baseline's actions.
Why it matters: It presents a concrete method for turning environment exploration into machine-checked mathematical proofs, with reported results on ARC-AGI-3 and formal impossibility claims.
Anthropic hiring manager Felix Rieseberg announces the end of traditional coding interviews, advising junior engineers to focus on business impact and rapid prototyping. Google DeepMind launches Genie 4, a real-time world model with audio and Gemini planning, while Cognition releases FrontierCode 2.0 benchmark. Axiom formalizes the Classification of Finite Simple Groups in Lean, and multiple model releases include Liquid LFM3 and GPT-6.1 Terra.
Anthropic used Claude to generate mathematical proofs that existing code is correct, producing roughly 180 million lines of proof code across 123 open-source libraries. The effort reportedly surfaced bugs that years of testing had missed. Anthropic's stated goal is to make AI-assisted formal verification as routine as standard testing.
OpenAI published new results on open problems in mathematics produced with an internal frontier model. The company also shared Lean proof formalizations and research details on GitHub. The release points to continued work on using frontier AI systems for mathematical discovery and formal verification.
