Signals from the deep
Stolen App Secrets: An Operator Is Signing Into Entra ID as Your Own Workloads
This content is available to DeepSeas customers. Sign in to view it.
Read report
About the portal
Research, investigations, and insights from DeepSeas experts.
We share findings from daily threat hunting, malware analysis, and incident investigations to help security teams understand:
Latest reports
Since 1 November 2026, 08:45 UTC, an unattributed actor has been signing in to Microsoft 365 and Azure tenants as the victims' own applications , using stolen app credentials that were still valid. No Microsoft vulnerability is involved. Once in, the actor harvests more secrets from Key Vault, searches Microsoft Graph, bulk-downloads SharePoint files and, through managed service provider (MSP) apps, reaches downstream customer tenants. Per public research, more than 33 app registrations at about 18 organizations authenticated successfully, and 27 organizations were reached through them.
Full analysis available to DeepSeas customers.
On 27 October 2026 TeamViewer published bulletin TV-2026-1011 , fixing three high-severity vulnerabilities in its remote-access clients, all resolved in version 15.83 . As of this writing there is no known exploitation in the wild — TeamViewer states it is unaware of any attacks, and none of the three CVEs appears in the CISA Known Exploited Vulnerabilities catalog — and no public weaponized proof-of-concept exploit has been identified. DeepSeas TIDE's recommended action is to update affected clients to 15.83 (or the corresponding legacy build) promptly, prioritizing the machines that initiate remote sessions.
Full analysis available to DeepSeas customers.
Between approximately 09:27 and 20:35 UTC on 29 October 2026, version 2.1.313 of the Claude Code editor extension ( anthropic.claude-code ), published under the official Anthropic namespace on the Open VSX registry and mirrored to Cursor's extension marketplace, carried a backdoor that was not developed by Anthropic. The injected loader runs as soon as the editor loads the extension, with no user interaction, and attempts to plant an attacker SSH key, open a reverse SSH tunnel to 134.209.82[.]225, and install persistence on macOS, Linux and Windows. Removing or updating the extension does not undo any of this. Any host that ran 2.1.313 must be checked for the svc-telemetry SSH key and the persistence artifacts listed in this report, whatever version is installed today.
Full analysis available to DeepSeas customers.
Between 20 and 27 October 2026, a single scripted clone farm registered lookalike domains for at least 33 healthcare, pharmaceutical and life-sciences brands, including Biogen, Becton Dickinson, Zoetis, Quest Diagnostics, Kenvue and Labcorp. The domains were registered in four bursts of seconds to minutes each. The farm grew from a root domain registered in May 2026. It now serves at least 330 corporate and retail lookalike hostnames from one web origin, at least 243 of its domains use one operator mail host, and it was still onboarding new targets on 28 October.
On 21 October 2026, Amazon Threat Intelligence reported seeing PHP2Shell exploitation in its MadPot honeypot network on 17 October — four days before CVE-2026-109618 and CVE-2026-109619 were publicly disclosed. DeepSeas TIDE independently observed the same activity and corroborates Amazon's central finding: the exploitation is real, and it began before disclosure. The identifiers were published on 21 October after an early leak of the embargoed detail, so no CVE existed to match against while the activity was recorded; the DeepSeas TIDE detections below are behavioural, which is why they predate the identifiers.
Full analysis available to DeepSeas customers.
PHP has released PHP 8.5.12, 8.4.27, 8.3.36 and 8.2.35 , fixing CVE-2026-109618 and CVE-2026-109619 ("php2shell"), two critical flaws (CVSS v4.0 9.2) that can allow unauthenticated remote code execution (RCE). PHP states that "there are no complete mitigations, updating is the only safe choice." Clients should update every PHP installation today, starting with Internet-facing servers.
Full analysis available to DeepSeas customers.
A daily briefing. Full analysis for DeepSeas customers.
20 November 2026
Friday, November 20, 2026
19 November 2026
Thursday, November 19, 2026
18 November 2026
Wednesday, November 18, 2026
16 November 2026
Monday, November 16, 2026
13 November 2026
Friday, November 13, 2026
DeepSeas expertise
Insights from an experienced team of threat hunters, reverse engineers, and analysts.
Our Threat Intelligence team draws on what we see across the DeepSeas client base, not just open-source feeds. That means faster, more relevant insight into which threats matter most.
Our knowledge base is continuously enriched with new adversaries, campaigns, TTPs, infrastructure, and attack patterns.
Research is turned into detection rules, IOCs, signatures, and other technical content.
We publish research, share findings, and contribute to the wider security community.
Services
24/7 MDR for lean security teams across mixed environments, built around existing security tools.
Rapid response to cyberattacks, from threat containment through recovery.
Penetration testing, attack simulations, and vulnerability assessments from our ethical hackers.
Threat data from our own research and other sources in one system for proactive defense.
Risk assessments, framework alignment, and compliance readiness built around your business.
Security strategy, program design, board reporting and roadmaps to reduce risk and strengthen security.
Stay informed
Full access to the latest reports, practical analysis, and defensive recommendations comes through our Threat Intelligence service.
Contact usPrefer to just follow new public reports?