Handshake has acquired Cleanlab in a deal aimed at strengthening AI research and data quality for training frontier models. In a CEO letter announcing the transaction, Cleanlab announced its team will focus on leading Handshake’s AI research efforts, with an emphasis on producing higher-quality training data and improving reliability across AI systems.
Cleanlab said Handshake is acquiring its latest research and technologies spanning confident learning, data-centric AI, and large language model (LLM) evaluation. The combined organization is framing the acquisition as a pairing of Handshake’s talent network with Cleanlab’s research capabilities to serve frontier AI labs and enterprise customers seeking better data quality and model trustworthiness.
Cleanlab also said its open-source data-centric AI package will remain available and will move to a more permissive Apache-2.0 license. The company highlighted adoption metrics including over one million downloads, “billions of AI datapoints cleaned,” and use by more than 100 Fortune-100 companies. For enterprise customers, Cleanlab emphasized that AI reliability challenges are intensifying and said the acquisition positions the combined company to address data quality “at the source,” upstream of model training.
In the letter, Cleanlab CEO and co-founder Curtis described learning during the M&A process that other AI data labs were sourcing human talent through Handshake’s network, which Cleanlab described as 20 million pre-vetted fellows. Cleanlab said Handshake’s leadership pivoted to help frontier labs access talent directly and that this approach helped reach a 100 million run rate in seven months. Cleanlab cast its role as adding the missing research component to partner with frontier AI labs, positioning the acquisition as creating a more fully integrated “data lab” offering.
Cleanlab, founded in 2021 by three MIT computer science PhDs, said it is trusted by more than 100 Fortune-500 organizations and is recognized as a Forbes AI 50 company. The company is backed by investors including Menlo Ventures, Bain Capital Ventures, TQ Ventures, Databricks Ventures, Samsung Ventures, and various angel investors.
