What's happening in email security: May 11 – May 17, 2026
DMARCbis is finally an RFC. Mail servers also had a rough week. Microsoft disclosed an actively exploited Exchange zero-click that fires when a user opens a crafted email in Outlook Web Access, Exim shipped a fix for a CVSS 9.8 remote code execution flaw triggered by a single TLS quirk, and a Belarus-aligned APT spent the spring running geofenced PDF phishing against Ukrainian government targets. If you run your own email infrastructure, you have patching and reading to do.
Crafted email triggers JavaScript in Outlook Web Access
CVE-2026-42897 is a cross-site scripting flaw in on-premises Exchange Server's OWA HTML rendering pipeline. The attacker sends a message whose HTML body slips JavaScript past OWA's sanitizer. When the recipient opens the message in their browser, the payload executes inside the OWA origin and inherits the user's authenticated session. From there it can read mail, send on the user's behalf, exfiltrate the session cookie, or call any OWA API against mailbox content the user can see. No attachment to click, no link to follow, just opening the message.
Microsoft hasn't published the sanitizer bypass technique or named the actors exploiting it. The fact that the official fix is a URL rewrite rule strongly suggests the payload sits in a URL-bearing attribute (an href or src that OWA renders without scrubbing) rather than in a script tag or event handler.
CISA added the CVE to the KEV catalog on May 15 and gave federal civilian agencies until May 29 to mitigate.
Exchange Server 2016, 2019, and Subscription Edition are all affected at every update level (CVSS 8.1). Exchange Online isn't affected. There's no patch yet, and Microsoft says updates are coming "in the future" for select cumulative updates only. ESU Period 1 customers won't get one at all.
The current mitigation is Microsoft's Emergency Mitigation Service, which applies a URL rewrite rule to neutralize the payload. It works, but it breaks calendar printing, blocks inline images in some messages, and disables OWA Light completely. Security teams are triaging which side effect their users will tolerate.
Several researchers are openly calling this Microsoft's strongest nudge yet toward the cloud, since Exchange Online customers are unaffected and on-prem patches look like they'll arrive slowly and selectively.
Exim ships emergency fix for CVSS 9.8 BDAT vulnerability
Exim 4.99.3 fixes CVE-2026-45185, a use-after-free in BDAT message body parsing. The exploit sends a TLS close notification followed by cleartext bytes during a CHUNKING SMTP session, which lets an attacker run code on the mail server. The researcher who found it, Federico Kirschbaum at XBOW Security Lab, said "almost no special configuration" is needed to trigger the bug.
Only GnuTLS-based Exim builds are affected. OpenSSL builds aren't. Versions 4.97 through 4.99.2 are vulnerable. If your distro packages Exim with GnuTLS (some Debian and Red Hat derivatives do), patch now. Exim runs a significant share of the world's MTAs, so check every mail relay you operate.
Ghostwriter uses geofenced PDFs to deliver Cobalt Strike to Ukrainian government
The Belarus-aligned group Ghostwriter (UNC1151, FrostyNeighbor, Storm-0257) has been running spear-phishing against Ukrainian military and government targets since March. The lure is a PDF impersonating Ukrtelecom. Victims who clicked received a RAR archive that dropped a JavaScript loader called PicassoLoader, which fetched Cobalt Strike Beacon.
The evasion layer is the interesting part. The PDFs were geofenced, so victims outside Ukrainian IP ranges received a benign document. After the loader ran, the system fingerprinted itself every 10 minutes and waited for an operator to manually approve final-stage payload delivery. ESET confirmed the campaign in parallel. Poland and Lithuania were also in scope.
Charities and airlines are the go-to impersonation targets right now
ESET's WeLiveSecurity team mapped a wave of scams riding Middle East conflict coverage. Two impersonation patterns stand out. The first is fake charity domains and lookalike sites soliciting donations for conflict relief. The second is airline impersonation, with scammers spoofing carriers running supposed refund or rebooking flows tied to flight disruption headlines. Government agency impersonation (visa processing, evacuation registration) shows up in the same campaigns, and credential harvesting is the consistent end goal. If you protect a charity, an airline, or any brand whose name keeps appearing in conflict-adjacent headlines, check your DMARC enforcement this week and watch for new lookalike registrations.
DMARCbis is officially an RFC
After years in draft, the IESG published DMARCbis as RFC 9989, with the aggregate reporting spec as [RFC 9990](https://www.rfc-editor.org/rfc/rfc9990.txt) and failure reporting as [RFC 9991](https://www.rfc-editor.org/rfc/rfc9991.txt). Together they obsolete RFC 7489 and RFC 9091. Existing records keep working since v=DMARC1 didn't change, but three tags are gone (pct, rf, ri), three are new (psd, np, t), and organizational domain discovery now uses a DNS Tree Walk instead of the Public Suffix List. Terminology shifts too. Policies are now Monitoring Mode or Enforcement Mode, and receivers are Report Consumers. The np tag is the practical headline, since it lets domain owners set a separate policy for non-existent subdomains and closes the subdomain spoofing gap that's been DMARC's most-abused weakness. Read [Red Sift's full guide on DMARCbis for more details](https://redsift.com/guides/dmarcbis-guide)
- FamousSparrow re-exploited Exchange three times against one Azerbaijani target
China-aligned APT with overlap to Earth Estries and Salt Typhoon ran three waves between December 2025 and February 2026, each time re-using the ProxyNotShell chain with a different backdoor after the victim remediated, a useful reminder that credential rotation and full forensic review after Exchange exploitation aren't optional.
- Kimsuky shifts to Rust and abuses VSCode tunneling
North Korean APT43 evolved PebbleDash with Rust-based loaders and uses legitimate tools (VSCode tunnels, DWAgent) for post-exploitation, still arriving via spear-phishing with quotation and job offer lures.
- Microsoft Patch Tuesday covers 138 vulnerabilities including Entra ID spoofing
Beyond the Exchange CVE, the May release includes a critical SSO Plugin auth bypass (CVE-2026-41103), a Netlogon RCE (CVE-2026-41089, CVSS 9.8), and an Azure Entra ID spoofing fix (CVE-2026-40379) worth tracking given Entra ID's role in M365 email authentication.
Want email security news in your inbox every week?
Subscribe here and stay up to date with what's happening in email security.




