Probably

Hypercritical, founded in 2023 in London, UK, uses a machine learning architecture combined with formal logic to generate rigorously verified control software for safety-critical sectors including automotive, aerospace, defense, and robotics, aiming to eliminate hallucinations and errors found in traditional large models when used in control code.

Resect AI, headquartered in Washougal, Washington, builds an accountability layer for enterprise AI. Its patented technology works in-stream, observing what happens inside large language models as they generate output, detecting and correcting the internal behaviors that lead to hallucinations rather than checking generated text after the fact, while producing an audit trail for enterprise compliance. Its flagship product, Resect NeuroWave, is described as a polygraph for neural networks, and the company has open-sourced Veritas fact-checking models (0.6B and 8B, built on Qwen3) on HuggingFace. Target industries include publishing, finance, healthcare, research and education.

WisdomAI, founded in 2023 in San Mateo, CA, USA, develops tools that improve AI training data quality to reduce hallucinations in large language models. The company claims ConocoPhillips, Cisco, and Descope as early customers, and serves clients using major cloud data storage services including Snowflake, Google BigQuery, Amazon Redshift, Databricks, and Postgres.

Gray Swan, founded in 2023 in Pittsburgh, USA (originating from Carnegie Mellon University), develops tools for testing and protecting AI models and agents against unsafe behaviors.

Maxim AI, founded in 2021 in San Francisco, USA, develops tools for evaluating and observing AI models.

AIMon, founded in 2022 in Palo Alto, USA, develops tools to help enterprises identify and correct errors in AI-generated content, such as factual inaccuracies or irrelevant information.