Security breach at AI aggregator exposes over 18 million files, including some very embarrassing chat records


Researchers found 22.7TB of exposed data from AI aggregator 302.ai, including chat logs and personal information. Leaked records contained user prompts, AI responses, metadata, and sensitive corporate documents. Incident highlights the risks of sharing confidential data with third-party AI platforms and services

Read more…
Source:  


Sign up for the Cyber Security Review Newsletter
The latest cyber security news and insights delivered right to your inbox


Related:

  • Security company faces backlash for waiting 12 months to disclose Palo Alto 0-day

    November 12, 2021

    There has been considerable debate within the cybersecurity community about Randori, a security firm that waited one year before disclosing a critical buffer overflow bug it discovered in Palo Alto Networks’ GlobalProtect VPN. The zero-day — which has a severity rating of 9.8 and was first reported by ZDNet — allows for unauthenticated, remote code execution ...

  • Mac Zero Day Targets Apple Devices in Hong Kong

    November 12, 2021

    Since at least late August, attackers have been using flaws in macOS and iOS – including in-the-wild use of what was then a zero-day flaw – to install a backdoor on the Apple devices of users who visited Hong Kong-based media and pro-democracy sites. This isn’t a finely targeted campaign, but it’s a sophisticated one. The ...

  • AMD reveals an EPYC 50 flaws – 23 of them rated High severity.

    November 12, 2021

    Microsoft may have given us a mere 55 CVEs to worry about on November’s Patch Tuesday, but AMD and Intel have topped that number with fixes for their products. AMD alone dropped 50 new CVEs on Thursday, 23 of them rated of “High” concern, meaning they’re rated at between 7.0 and 8.9 on the Common Vulnerability ...

  • QAKBOT Loader Returns With New Techniques and Tools

    November 12, 2021

    QAKBOT is a prevalent information-stealing malware that was first discovered in 2007. In recent years, its detection has become a precursor to many critical and widespread ransomware attacks. It has been identified as a key “malware installation-as-a-service” botnet that enables many of today’s campaigns. Toward the end of September 2021, we noted that QAKBOT operators resumed ...

  • BotenaGo botnet targets millions of IoT devices with 33 exploits

    November 11, 2021

    The new BotenaGo malware botnet has been discovered using over thirty exploits to attack millions of routers and IoT devices. BotenaGo was written in Golang (Go), which has been exploding in popularity in recent years, with malware authors loving it for making payloads that are harder to detect and reverse engineer. In the case of BotenaGo, only ...

  • TeamTNT Upgrades Arsenal, Refines Focus on Kubernetes and GPU Environments

    November 11, 2021

    In previous entries, we described how the hacking group TeamTNT targeted unsecured Redis instances, exposed Docker APIs, and vulnerable Kubernetes clusters in order to deploy cryptocurrency-mining payloads and credential stealers. TeamTNT was one of the first cybercriminal groups to focus on cloud service providers (CSPs), specifically the metadata stored on elastic computing instances being run ...