AI startup Anthropic, valued at $18 billion, confirmed a data leak just two days after the Federal Trade Commission (FTC) initiated an inquiry into generative AI investments involving five major firms, including Anthropic.

On January 22, Anthropic, founded by ex-OpenAI siblings Dario and Daniela Amodei, discovered that a contractor inadvertently shared a file containing non-sensitive customer information with a third party. The leaked data included a subset of customer names and open credit balances from the year 2023. Anthropic assured that this incident was caused by human error and not related to their in-house AI systems.

Affected customers have been promptly notified and provided with relevant guidance, according to a spokesperson from Anthropic. However, the company refrained from commenting on the ongoing FTC probe into its $4 billion investment from Amazon in September 2023. The FTC’s investigation covers similar partnerships involving Alphabet, Microsoft, and OpenAI, seeking information on agreements, strategic rationales, and practical implications, including decisions on product releases, governance, and oversight rights.

The leaked information, acknowledged by Anthropic as an isolated incident, has raised concerns among enterprises deploying third-party large language models. Anthropic’s model, Claude, has been made available to safety- and societal impact-minded researchers through a special research access program since 2022, aligning with the company’s public benefit mission.

Windows Report recently obtained an email from Anthropic to affected customers acknowledging the inadvertent data leak. The email clarified that the leaked information did not include sensitive personal data, such as banking or payment information. Despite characterizing the incident as an isolated error, Anthropic urged customers to remain vigilant against potential phishing attempts or suspicious communications.

VentureBeat’s inquiry revealed that only a subset of Anthropic’s users were impacted by the leak, although the specific number was not disclosed. This data breach, occurring amid a surge in breaches attributed to human error, highlights the concerns of enterprises entrusting proprietary data to third-party large language models.

Anthropic, valued at $18.4 billion and experiencing rapid growth since its 2021 inception, faces scrutiny from the FTC regarding its strategic partnerships with Amazon and Google. The agency’s 6(b) orders seek detailed information on multi-billion-dollar relationships, including the extended partnerships announced by Microsoft and OpenAI, Amazon and Anthropic’s collaboration, and Google’s expanded AI partnership with Anthropic.

As regulatory bodies like the FTC intensify their focus on AI partnerships, the leak underscores the challenges faced by companies navigating complex collaborations in the rapidly evolving AI landscape. Anthropic’s close relationships with AWS and Google, as well as the ongoing investigations by regulatory bodies, further emphasize the need for transparency and vigilance in AI partnerships to ensure fair competition and data security.