Researchers at Noma Labs showed that GitHub's new Agentic Workflows, which let an AI agent read issues and act on repositories automatically, can be tricked into leaking private code through nothing more than a public issue. The technique, GitLost, is indirect prompt injection: an attacker opens an ordinary-looking issue in an organization's public repository, buries plain-English instructions in it, and the agent, which often holds a token with read access across the org's repositories, follows them, fetches files from a private repo, and posts the contents in a public comment. No credentials, coding, or write access are needed. GitHub was notified, but researchers frame it as an architectural weakness.
The ShinyHunters extortion group has now published the Charter Communications data it stole, after the telecom giant apparently refused to pay. Earlier reporting put the breach at 4.9 million HIBP-confirmed unique accounts; ShinyHunters' leak is described as potentially impacting up to 5 million customers. Charter is one of the largest US telecoms, providing internet, cable, mobile, and phone services to residential and business customers under the Spectrum brand. The data was originally exfiltrated via voice-phishing of a Microsoft Entra account on April 1 and a Salesforce export. With the data now public rather than merely claimed, the phishing and identity-theft risk to affected customers rises sharply.
An Amazon S3 bucket simply named 'tabiq' was left open to anyone who knew the name, exposing over a million passports, driver's licenses, and identity-verification selfies submitted by hotel guests worldwide. The platform, run by Japanese operator Reqrea, handles digital check-in. Researcher Anurag Sen found the bucket and notified TechCrunch and JPCERT; the bucket has since been locked down. Reqrea says the exposed files date from early 2020 through May 2026 and that it does not yet know how the bucket became public. The company is still reviewing access logs to determine whether anyone else accessed the data.
TrustStrike Labs