AI Security & Cybersecurity

What Happened Unit 42 documented a voice-phishing (vishing) campaign — dubbed "Spring Ring" — that abused Microsoft Teams calling features to impersonate trusted internal contacts, socially engineer employees, deliver malware, and escalate to domain controllers, enabling broad network compromise [1]. The campaign leveraged platform-native trust signals (caller identity, in-app presence) to bypass traditional email-centric defenses…

What Happened Microsoft observed a multistage intrusion campaign (TerminalFix / ClickFix variant) that combined social engineering, signed‑binary abuse, steganography and a Python‑based reverse tunnel to enable silent pivoting and reconnaissance inside target networks [1]. Initial access: victims were lured to paste a malicious PowerShell command from a fake Cloudflare Turnstile CAPTCHA; that command…

What Happened Unit 42 introduced a diagnostic called perturbation probing and used it to show that safety refusals in large language models are often concentrated in a thin, localized neural layer rather than distributed across the model. The practical takeaway is that internal model defenses (the model's own refusal behavior) can be highly brittle: small…

What Happened Microsoft announced expanded security and governance controls aimed at organizations deploying AI agents and integrating third‑party telemetry. Key points include: Extended managed detection and response: Microsoft Defender Experts MDR (P2) now ingests third‑party data through Microsoft Sentinel, enabling 24/7 MDR and threat hunting across non‑Microsoft sources such as Palo Alto Networks,…

What Happened Two converging trends in 2026 make AI infrastructure a uniquely valuable attacker target: (1) adversaries are compromising AI gateways, retrieval/orchestration platforms and runtimes to steal provider credentials, establish persistence and monetize compute; and (2) advanced autonomous agents can discover and weaponize zero‑days to escape virtual machines and operate as APTs. Gateways and orchestration…

What Happened Security research and vendor reports show two converging trends that change defensive priorities. First, the traditional disclosure → assess → patch cycle is no longer fast enough: attackers and AI tools compress exploitation timelines to hours while defenders still need days or weeks to validate and deploy fixes. That creates a widening asymmetric…

What Happened Security research and incident response teams report a clear shift in attacker focus: instead of primarily exploiting production application code, adversaries increasingly target the software development lifecycle (SDLC) — CI/CD pipelines, developer tools, artifact registries, and model training pipelines. Compromises in these areas let attackers insert malicious dependencies, steal secrets, backdoor models, or…

What Happened Two themes dominate recent defensive research: attackers are exploiting trusted collaboration and identity channels to harvest credentials and tokens, and enterprise runtime gaps across cloud and Kubernetes are increasing exposure for workloads — including AI models and agentic components. Unit 42 documented identity abuse via trusted collaboration channels where impersonation, malicious…

What Happened Industry research and vendor benchmarking show a clear shift: vulnerabilities become critical only when they are exposed and combined with misconfiguration, over‑permissioned identities and live runtime activity, so defenders are moving security controls down into runtime. Kubernetes now runs in roughly 82% of production environments, driving adoption of a single runtime security model…

What Happened Security teams continue to see credential compromises used as the primary vector for large-scale cloud intrusions and downstream attacks on AI systems. In one recent instance, threat actor "TheHatman" claimed to have exfiltrated a large volume of credentials from Microsoft Entra tenants; Unit 42 published an updated mitigation brief with detection, containment and…

What Happened Two recent pieces of defensive and offensive research illustrate threats that cross messaging, IoT and AI infrastructure: Trail of Bits built and runs one of Signal’s independent auditors that enforces Automatic Key Verification by signing Merkle‑tree heads and limiting a server’s ability to present split views to clients to seven days [1]. Separately,…

What Happened Decentralized C2 using blockchains (Aeternum) Security researchers analyzed Aeternum, a botnet loader that implements command-and-control (C2) and communication channels via smart contracts on the Polygon blockchain. By putting C2 metadata on-chain and using public RPC endpoints, operators gain persistence, redundancy and increased difficulty of takedown compared with traditional centralized C2 infrastructure [1]. Rust-based,…