Device code phishing abuses Microsoft's OAuth 2.0 device authorization grant flow, allowing attackers to generate device codes and trick users into authorizing them on legitimate Microsoft login pages. This bypasses MFA by stealing valid OAuth tokens directly. The attack has surged recently, with campaigns like Storm-2372 targeting governments and NGOs, and phishing-as-a-service platforms like EvilTokens and Kali365 now offering it as a commodity service. Key mitigations include enabling Microsoft Entra Conditional Access policies to block device code flow, monitoring new device registrations in Entra, watching for suspicious token exchanges in logs, and disabling device code flow where operationally feasible. Notably, around 25% of organizations that have paid for Conditional Access have not yet configured it, leaving a significant gap.
Nguồn: https://www.huntress.com/blog/tradecraft-tuesday-device-code-phishing-explained. 8sync News chỉ tóm tắt và dẫn link; bản quyền nội dung thuộc tác giả và nguồn gốc.
LastPass xác nhận dữ liệu khách hàng trong môi trường Salesforce bị truy cập sau cuộc tấn công chuỗi cung ứng nhằm vào Klue hôm 12/6. Nhóm tống tiền Icarus đã xâm nhập hạ tầng Klue bằng thông tin đăng nhập cũ, đánh cắp token OAuth kết nối Klue với Salesforce của khách hàng. Dữ liệu bị lộ bao gồm tên, số điện thoại, email, địa chỉ, thông tin hỗ trợ và dữ liệu CRM. LastPass cho biết sản phẩm cốt lõi, dịch vụ và kho dữ liệu khách hàng không bị ảnh hưởng.
Lập trình viên nên đọc bài này để hiểu rõ về cách tấn công supply chain attack hoạt động như thế nào, từ đó nâng cao kiến thức bảo mật cho các ứng dụng và hệ thống của mình, đặc biệt là khi sử dụng các dịch vụ cloud như Salesforce.
Huntress discloses it was among multiple victims of a supply chain attack targeting Klue, a market intelligence platform. The threat actor, dubbed Icarus, compromised Klue's backend systems on June 11, 2026, injecting code to steal OAuth tokens used by Klue's customers to connect their CRM tools. This allowed the attacker to directly query and exfiltrate Salesforce data from Huntress and other companies including Recorded Future, Tanium, and Jamf. The stolen Huntress data includes business contact info, pricing, subscription details, and sales communications — no product telemetry, passwords, or payment data was affected. Huntress shares IOCs (IP addresses, User-Agent strings), threat actor attribution details linking to the Icarus extortion group, and five recommended investigation steps for other potentially impacted organizations. The post is being updated in real time as the situation evolves, with a secondary unauthorized party also claiming access to breach data as of June 24.
Market research company Klue, breached on June 12, is communicating with the hacking group Icarus and believes they are deleting stolen customer data. However, a second unnamed hacker group has emerged, claiming to have obtained Klue's customer data from Icarus and threatening to leak it unless a ransom is paid. The second group alleges 195 affected Klue customers and claims Klue paid Icarus. Klue advises customers contacted by this second group to demand a data sample as proof before taking any action. The original breach involved a 2022 third-party credential that was never revoked, which attackers used to steal OAuth tokens and access customer clouds and databases. Affected companies include Gong, Jamf, HackerOne, Huntress, LastPass, Snyk, and others.
AI readiness goes beyond choosing smart models — it requires mature data, governance, and integration foundations. For APIs specifically, this means shifting from human-readable interfaces to machine-consumable ones that AI agents can reliably discover, call, and recover from. Traditional APIs fail in agentic systems due to poor documentation, inconsistent error handling, non-idempotent behavior, human-driven authentication, oversized payloads, and infrastructure fragility. To become AI-ready, APIs should adopt strongly typed OpenAPI schemas, concrete examples, structured JSON error objects, idempotent endpoints, strict pagination, and machine-friendly auth like OAuth 2.0 client credentials. These upgrades enable enterprises to scale agentic architectures safely without risking compliance failures, cost overruns, or security exposures.
LastPass is notifying customers that personal data — including names, emails, phone numbers, and support case contents — was stolen via a supply chain breach at Klue, a competitive intelligence vendor. Hackers obtained OAuth tokens Klue held on behalf of enterprise clients, using them to access LastPass's Salesforce environment. The breach did not compromise LastPass's own infrastructure or encrypted password vaults. A hacking and extortion group called Icarus claimed responsibility and is threatening to release the data unless ransoms are paid. Other affected companies include HackerOne, Recorded Future, Snyk, and Huntress. The incident underscores a structural risk: when a single vendor holds OAuth tokens for dozens of enterprise customers, compromising that vendor grants simultaneous access to all of them. LastPass has revoked the tokens and is advising customers to watch for phishing attempts leveraging the stolen data.
Modern phishing attacks have evolved far beyond misspelled domains and fake login pages. Attackers now use ClickFix (tricking users into running malicious terminal commands via fake CAPTCHAs), Browser-in-the-Browser (BitB) attacks that render convincing fake browser windows inside real pages, OAuth consent phishing (ConsentFix) that steals authorization codes without ever asking for a password, device code phishing that abuses legitimate OAuth flows to authorize attacker devices, and fake video conference overlays that prompt malware downloads under the guise of driver updates. Each technique systematically eliminates traditional red flags, weaponizing users' own habits and trusted infrastructure against them. Huntress SAT offers simulated scenarios replicating these exact tactics to build behavioral muscle memory before users encounter the real thing.
Identity-Chained Authorization is a security pattern for AI agents that ensures every action is executed under a real human's identity rather than a shared service account. Using Auth0 Token Vault and durable workflows (Vercel WDK), the pattern stores per-engineer OAuth tokens during onboarding, then performs RFC 8693 token exchanges at the exact moment of human approval — never before. The LLM's tool schemas contain zero credential parameters; identity is resolved by the harness independently. The architecture handles token expiry in long-running workflows via a nested interrupt pattern: if a refresh token is stale, the workflow suspends again to prompt re-authorization before retrying. Three guarantees result: the model never sees a credential, tokens are always fresh at execution time, and every GitHub action carries a verifiable human author for a real audit trail.
AI agents are becoming the fastest-growing class of enterprise identities, yet most organizations lack a governance plan for them. Each agent should have its own distinct identity rather than borrowing human credentials or generic service accounts. Authentication relies on OAuth 2.0: use the on-behalf-of (OBO) flow when an agent acts for a user, and client credentials when acting autonomously. Short-lived federated credentials should replace long-lived secrets, and tokens should be scoped per resource. For high-risk actions, out-of-band human approval (e.g., hardware-key-backed CIBA flows) creates a verifiable accountability chain. Platforms like Microsoft Entra Agent ID, Okta, and CyberArk are building dedicated agent identity tooling. Best practices include assigning every agent a named owner with a lifecycle, enforcing least privilege, and continuous runtime audit logging.