DeepSeek Data Leak Exposes Over 1 Million Sensitive Log Streams

A new attack vector, dubbed ShadowLeak, allows hackers to invisibly steal emails from users who integrate AI agents like ChatGPT with their email inboxes. The attack exploits the lack of visibility into AI processing on cloud infrastructure, making it undetectable to the user. The vulnerability was discovered by Radware and reported to OpenAI, which addressed it in August 2025. The attack involves embedding malicious code in emails, which the AI agent processes and acts upon without user awareness. The attack leverages an indirect prompt injection hidden in email HTML, using techniques like tiny fonts, white-on-white text, and layout tricks to remain undetected by the user. The attack can be extended to any connector that ChatGPT supports, including Box, Dropbox, GitHub, Google Drive, HubSpot, Microsoft Outlook, Notion, or SharePoint. The ShadowLeak attack targets users who connect AI agents to their email inboxes, such as those using ChatGPT with Gmail. The attack is non-detectable and leaves no trace on the user's network. The exploit involves embedding malicious code in emails, which the AI agent processes and acts upon, exfiltrating sensitive data to an attacker-controlled server. OpenAI acknowledged and fixed the issue in August 2025, but the exact details of the fix remain unclear. The exfiltration in ShadowLeak occurs directly within OpenAI's cloud environment, bypassing traditional security controls.

The RaccoonO365 phishing network, a financially motivated threat group, was disrupted by Microsoft's Digital Crimes Unit (DCU) and Cloudflare. The operation, executed through a court order in the Southern District of New York, seized 338 domains used by the group since July 2024. The network targeted over 2,300 organizations in 94 countries, including at least 20 U.S. healthcare entities, and stole over 5,000 Microsoft 365 credentials. The RaccoonO365 network operated as a phishing-as-a-service (PhaaS) toolkit, marketed to cybercriminals via a subscription model on a private Telegram channel. The group used legitimate tools like Cloudflare Turnstile and Workers scripts to protect their phishing pages, making detection more challenging. The mastermind behind RaccoonO365 is believed to be Joshua Ogundipe, who received over $100,000 in cryptocurrency payments. The group is also suspected to collaborate with Russian-speaking cybercriminals. Cloudflare executed a three-day 'rugpull' against RaccoonO365, banning all identified domains, placing interstitial 'phish warning' pages, terminating associated Workers scripts, and suspending user accounts to prevent re-registration.

A vulnerability in the Cursor AI-powered Integrated Development Environment (IDE) allows automatic execution of tasks in malicious repositories upon opening. This flaw can be exploited to drop malware, hijack developer environments, or steal credentials and API tokens. The issue arises from Cursor disabling the Workspace Trust feature from Visual Studio Code (VS Code), which blocks automatic execution of tasks without explicit consent. This default behavior can be exploited by adding a malicious .vscode/tasks.json file in a publicly shared repository. The flaw affects Cursor's one million users who generate over a billion lines of code daily. The flaw can be exploited to leak sensitive credentials, modify files, or serve as a vector for broader system compromise, placing Cursor users at significant risk from supply-chain attacks. Cursor has decided not to fix the issue, citing the need to maintain AI and other features that depend on the autorun behavior. Users are advised to enable Workspace Trust manually or use a basic text editor for unknown projects.

Chinese state-sponsored Advanced Persistent Threat (APT) actors, specifically the Salt Typhoon group and a newly identified group named RedNovember, have been conducting sustained campaigns to compromise critical infrastructure networks worldwide. The campaigns aim to gain long-term access to telecommunications, government, transportation, lodging, and military networks. This activity has been detailed in a joint advisory by CISA, NSA, FBI, and international partners, including Canada, Australia, New Zealand, the UK, Czech Republic, Finland, Germany, Italy, Japan, the Netherlands, Poland, and Spain. The advisory provides intelligence on tactics used by these actors and recommends mitigations to strengthen defenses. The Czech Republic's National Cyber and Information Security Agency (NUKIB) has issued a warning instructing critical infrastructure organizations to avoid using Chinese technology or transferring user data to servers located in China. The agency has re-evaluated its risk estimate of significant disruptions caused by China, now assessing it at a 'High' level. The NUKIB has confirmed malicious activities of Chinese cyber-actors targeting the Czech Republic, including a recent APT31 campaign targeting the Czech Ministry of Foreign Affairs. The advisory highlights concerns over the transfer of system and user data to China, potentially misused by state, military, or political interests. The Czech government previously accused China of targeting its critical infrastructure through APT 31, an allegation denied by the PRC but condemned by the US, EU, and NATO. The advisory suggests that individuals and organizations consider restricting or prohibiting the use of products and services that transfer data to China. The campaign has targeted at least 600 organizations across 80 countries, including 200 in the U.S. The threat actors have exploited vulnerabilities in Cisco, Ivanti, and Palo Alto Networks devices to gain initial access and have modified routers to maintain persistent access and pivot into other networks. The advisory also notes that the APT actors may target other devices such as Fortinet firewalls, Juniper firewalls, Microsoft Exchange, Nokia routers and switches, Sierra Wireless devices, Sonicwall firewalls, etc. RedNovember has targeted perimeter appliances of high-profile organizations globally, including defense and aerospace organizations, space organizations, and law firms. The group has breached at least two U.S. defense contractors, a European engine manufacturer, and a trade-focused intergovernmental cooperation body in Southeast Asia. RedNovember has used the Go-based backdoor Pantegana and Cobalt Strike as part of its intrusions, along with the Spark RAT and LESLIELOADER. The group has also used VPN services like ExpressVPN and Warp VPN to administer and connect to servers used for exploitation and communication.

The threat actor, tracked as UNC6395 by Google and GRUB1 by Cloudflare, exploited OAuth tokens associated with the Drift AI chat agent to breach Salesloft and steal data from Salesforce customer instances. The campaign, active from August 8 to at least August 18, 2025, targeted over 700 organizations, including Workiva and Stellantis, and impacted all integrations connected to the Drift platform, not just Salesforce. The attackers exported large volumes of data, including credentials for AWS, passwords, and Snowflake access tokens. Zscaler, Palo Alto Networks, Cloudflare, and Workiva reported data breaches after threat actors accessed their Salesforce instances via compromised Salesloft Drift credentials, exposing customer information. The breach began with the compromise of Salesloft's GitHub account, accessed by UNC6395 from March to June 2025. The threat actor accessed multiple repositories, added a guest user, and established workflows. Reconnaissance activities occurred in the Salesloft and Drift application environments between March and June 2025. The attackers accessed Drift's AWS environment and obtained OAuth tokens for Drift customers' technology integrations. Salesloft isolated the Drift infrastructure, application, and code, and took the application offline on September 5, 2025. Salesloft rotated credentials in the Salesloft environment and hardened it with improved segmentation controls. Salesloft recommends that all third-party applications integrated with Drift via API key revoke the existing key. Salesforce restored the integration with the Salesloft platform on September 7, 2025, except for the Drift app, which remains disabled. Salesloft and Salesforce have taken steps to mitigate the breach, including revoking tokens and removing the Drift application from AppExchange. The breach highlights the risks associated with third-party integrations and the potential for supply chain attacks. UNC6395 demonstrated operational discipline, querying and exporting data methodically, and attempting to cover their tracks by deleting query jobs. The targeted organizations included security and technology companies, suggesting a broader strategy to infiltrate vendors and service providers. The campaign is limited to Salesloft customers who integrate their own solutions with the Salesforce service. There is no evidence that the breaches directly impacted Google Cloud customers, though any of them that use Salesloft Drift should review their Salesforce objects for any Google Cloud Platform service account keys. The threat group ShinyHunters and Scattered Spider claimed responsibility for many of those attacks, and vishing attacks have been cited as the means of compromise. Google disclosed that UNC6040 breached one of its Salesforce instances using these tactics. The UNC6395 Salesloft Drift activity is separate from the vishing attacks attributed to UNC6040. Okta successfully defended against a potential breach by enforcing inbound IP restrictions, securing tokens with DPoP, and using the IPSIE framework. Okta recommends that organizations demand IPSIE integration from application vendors and implement an identity security fabric. Palo Alto Networks' Unit 42 advised organizations to conduct immediate log reviews for signs of compromise and rotate exposed credentials. Okta suggests reducing the blast radius of a single entity breach by constraining token use by IP and client and ensuring granular permissions for M2M integrations. The FBI has issued a FLASH alert warning that two threat clusters, tracked as UNC6040 and UNC6395, are compromising organizations' Salesforce environments to steal data and extort victims. UNC6040 is a threat actor that specializes in voice phishing or vishing and recently was observed using social engineering to pose as IT support staff to get into Salesforce environments. UNC6395 is best known for using stolen OAuth tokens from Salesloft's Drift application, which has a Salesforce integration, to steal sensitive data from hundreds of Salesforce environments earlier this year. The FBI's latest advisory provides additional context into the technical aspects of the threat campaigns, particularly UNC6040's activity, which began last fall. The advisory also includes indicators of compromise, including IP addresses and URLs associated with the two campaigns.