Full Report
This week starts small. A token leaks. A bad package slips in. A login trick works. An old tool shows up again. At first, it feels like the usual mess. Then you see the pattern: attackers are not always breaking in. They are using the parts we already trust. That is what makes it worrying. The danger is in normal things now - updates, apps, cloud buttons, support chats, trusted accounts. AI
Analysis Summary
# Morning News Roll-up May 21, 2026
## Overview
The current threat landscape is shifting from traditional "breaking and entering" to the exploitation of trusted components. Attackers are increasingly leveraging "normal" elements—updates, cloud functions, support chats, and trusted account infrastructures—to facilitate intrusions. AI is being utilized primarily as a force multiplier to increase the velocity and scale of these identity-based and social engineering attacks.
## Top Stories
### Pwn2Own Berlin 2026 Reveals 47 Zero-Days
- Summary: Security researchers exploited 47 zero-day vulnerabilities in major platforms including Windows 11, VMware, and Microsoft Exchange, highlighting significant remaining gaps in widely trusted enterprise software.
- Source: hxxps://www[.]zerodayinitiative[.]com/blog/2026/5/16/pwn2own-berlin-2026-day-three-results-and-master-of-pwn
### U.K. NCSC Warns of Agentic AI Risks
- Summary: The National Cyber Security Centre issued guidance on "Agentic AI," warning that over-privileged or poorly designed AI agents can turn a single failure into a major incident if not properly governed.
- Source: hxxps://www[.]ncsc[.]gov[.]uk/blogs/thinking-carefully-before-adopting-agentic-ai
### Poland Urges Move Away from Signal Due to Support Impersonation
- Summary: Polish authorities are directing officials to use a domestic encrypted messenger (mSzyfr) following a rise in APT-led social engineering attacks where hackers impersonate Signal support to hijack accounts.
- Source: hxxps://www[.]gov[.]pl/web/baza-wiedzy/rekomendacja-pelnomocnika-rzadu-ds-cyberbezpieczenstwa-dotyczaca-komunikatora-signal
# Main Topic
The Exploitation of Trusted Identity and "Normal" Infrastructure
## Key Points
- Attackers are moving away from traditional exploits toward "using the parts we already trust," such as cloud buttons, support chats, and legitimate app updates.
- Identity is the primary perimeter; the leakage of single tokens or successful "login tricks" are providing initial access.
- AI is being deployed to accelerate the "brute-forcing" of social engineering and discovery, rather than creating fundamentally new attack types.
- A significant metric for modern CISOs is now "Time-to-Revoke" (TTR) for compromised machine and human identities.
## Threat Actors
- **Advanced Persistent Threat (APT) Groups:** Unspecified groups targeting government officials via messenger app impersonation.
- **Financial Fraudsters:** Groups utilizing scam kits and social engineering; notably, Dutch authorities are tracking at least 100 suspects in this category.
- **DEVCORE, STARLabs SG:** (Security Researchers) Highlighted the continued vulnerability of core enterprise systems (SharePoint, Exchange).
## TTPs
- **Support Impersonation:** Threat actors impersonate platform support (e.g., Signal support) to gain account control.
- **Social Engineering:** High-pressure tactics to trick users into providing credentials or tokens.
- **Token Theft:** Leveraging leaked or phished session tokens to bypass MFA.
- **Agentic AI Exploitation:** Capitalizing on over-privileged AI agents to execute unauthorized enterprise actions.
## Affected Systems
- **Communication Platforms:** Signal (Target of impersonation campaigns).
- **Enterprise Software:** Microsoft SharePoint, Exchange, Windows 11, VMware (Target of zero-day exploits).
- **Cloud Infrastructure:** "Machine identities" and non-human identities serving as a primary governance gap.
## Mitigations
- **Identity Governance:** Implement strict "Time-to-Revoke" protocols for all credentials and session tokens.
- **AI Security Controls:** Apply the principle of least privilege to "Agentic AI" tools to prevent autonomous escalations.
- **Verification Protocols:** Establish out-of-band methods for verifying support communications from encrypted messaging services.
- **Indigenous Infrastructure:** Moving highly sensitive government communications to dedicated, sovereign encrypted platforms (e.g., mSzyfr).
## Conclusion
The threat landscape has evolved into a battle over identity and trust. Defensive strategies must prioritize the rapid revocation of access (TTR) and the implementation of zero-trust principles for non-human identities and AI agents. Organizations should expect social engineering to become more sophisticated and high-volume as attackers integrate AI to automate the initial stages of the kill chain.