Tigerfish, a San Francisco-based transcription firm whose customers include police departments, prosecutors offices and Silicon Valley technology companies, has closed a security lapse that left a large volume of internal data exposed on the open web. The material included completed transcriptions along with the original audio and video that customers had submitted for processing. Police interview transcripts, body-camera footage and customer invoices were among the files that could be reached online.
The files sat on an Amazon-hosted S3 cloud server that was publicly listed and reachable through an easily guessed web address, according to Anurag Sen, the security researcher who found them. No password was required to access the contents. Sen, 30, who works from his home in Indonesia, spotted the data in October and contacted the company, but received no reply, so he passed details to reporters in an effort to get the exposure addressed.
After being contacted about the findings, Tigerfish restricted access to the storage bucket. Company president Adam Goldberg said in an email that all public access to the identified bucket had been blocked, including directory listings and individual files, and that the company had confirmed its other S3 buckets were also closed to public access. Goldberg said Tigerfish is continuing to review and strengthen its storage security more broadly. The company said Sen's initial email had been caught in its spam filter, and it did not answer a follow-up email asking whether customers had been notified.
The incident illustrates how a simple misconfiguration can put sensitive information online. Sen said most of the exposures he finds trace back to basic human error rather than sophisticated attacks. He described an engineer standing up a test cloud database on a Friday, turning off authentication to test an API, and forgetting to close it before the weekend. He argued that the real risk is sensitive systems left open to the public internet without a password, even as attention focuses on shadowy hackers.
Sen has spent the past decade hunting for exposed datasets and alerting the organizations behind them, work he describes as an unpaid air-traffic controller. His past finds include sensitive U.S. military emails spilling from a Department of Defense cloud server, databases exposing customer records, the inner workings of a national license plate surveillance system, and public systems that sent out SMS login codes used around the world. He said the modern web runs on an assumption that data is kept secure, while his findings show how fragile that arrangement is behind the scenes.
The discovery comes as AI tools make it easier to uncover weaknesses, and as AI-assisted development pushes unchecked code onto the web. Amazon has taken steps to make it harder to make S3 buckets public without clear multistep warnings, and newer platforms such as Lovable and Supabase have made changes aimed at preventing user mistakes. Tigerfish did not respond to a follow-up email.
More hardware news from TechManNews.





