SentinelOne

🔥 🟣 Purple AI is here and now generally available! To learn more about the industry’s leading AI security analyst, watch the video below. 👉 Ready to transform your security operations? Get a demo: sentinelone.com/platform/purpl…

Law enforcement took down a Russian-linked hosting network, a ransomware group escalated to dispatching physical operatives for data extortion, and a massive supply chain campaign targets developer environments and AI tools. This is the Good, Bad & Ugly. ⬇️ ✅ GOOD - Dutch authorities dismantled Stark Industries, seizing 800 servers used to enable pro-Russian DDoS and disinformation campaigns. - A Romanian hacker received a 56-month federal prison sentence for breaching the Oregon state government network and stealing PII. ⚠️ BAD - The FBI warns that Silent Ransom Group is targeting U.S. legal and financial institutions with in-person data extortion schemes. - Attackers use typosquatted helpdesk domains, and if remote access fails, deliberately dispatch physical operatives to manually insert USBs into company computers. - The attackers then harass employees and clients by phone to force financial negotiations under the threat of leaking proprietary data. 🤢 UGLY - Security researchers uncovered TrapDoor, a massive supply chain attack spreading credential-stealing malware across npm, PyPI, and Crates.io. - The campaign leverages registry-specific execution methods to harvest sensitive developer secrets, cloud credentials, and cryptocurrency wallets. - Threat actors uniquely implant poisoned files designed to trick AI coding assistants into autonomously executing malicious security scans. Full breakdown → s1.ai/GBU9-Wk22

From day one, SentinelOne was architected to stop novel, machine-speed threats. We were purpose built to be a Leader in the AI era. For the sixth consecutive year, Gartner has named SentinelOne a Leader in the Gartner® Magic Quadrant™ for Endpoint Protection. What's driving the recognition: ✅ Autonomous detection and response at machine speed ✅ Unified visibility across endpoint, identity, cloud, and AI ✅ AI usage control through the Prompt Security acquisition ✅ AI-native from day one — not retrofitted 📖 Read the full report: s1.ai/GrtnrMQ26

~50% of SentinelOne's ARR now comes from emerging solutions. That's what platform expansion looks like. This quarter, our emerging solutions — AI, Data, Cloud, and more — reached half of our total ARR, alongside record net new ARR growth and the launch of Purple AI Auto-Investigations. 📈 Q1 FY2027 highlights: $1.163B ARR — +23% YoY $277M Revenue — +21% YoY Record net new ARR growth 4% Operating Margin (non-GAAP) — ~550 bps improvement YoY 22% Adjusted FCF Margin (non-GAAP) — ~230 bps improvement YoY $0.04 EPS (non-GAAP) — +83% YoY ~50% of ARR from Emerging Solutions Securing modern enterprises requires machine-speed defense — and infrastructure built for what's next, not retrofitted for it. Thank you to our customers, partners, and Sentinels. 🔗 Read the press release: s1.ai/Q1-27-PR 🎧 Listen to the call: s1.ai/Q1-27-Ern

From day one, SentinelOne was architected to stop novel, machine-speed threats. We were purpose built to be a Leader in the AI era. For the sixth consecutive year, Gartner has named SentinelOne a Leader in the Gartner® Magic Quadrant™ for Endpoint Protection Platforms. Autonomous detection. Machine-speed response. Built for this moment. 📖 Read the report: s1.ai/GartnerMQ-2026

Law enforcement took down cybercrime operations and attacker-controlled VPNs, a multi-brand spoofing macOS infostealer slips past OS mitigations, and Microsoft Defender faces zero-day exploits. This is the Good, Bad & Ugly. ⬇️ ✅ GOOD - INTERPOL coordinates 201 arrests across the MENA region in a massive cybercrime sweep. - Ukraine identifies an infostealer operator behind 28,000 stolen credentials. - International police seize a dedicated commercial VPN provider used explicitly for ransomware exfiltration. ⚠️ BAD - SentinelOne identifies "SHub Reaper," a new macOS stealer variant that impersonates Apple, Google, and Microsoft in a single attack chain. - Bypasses Apple's new Terminal security mitigations by abusing the native AppleScript URL handler. Acts as both a smash-and-grab credential thief and a persistent backdoor for remote access. 🤢 UGLY - Microsoft warns that attackers are actively exploiting two new Windows Defender zero-day vulnerabilities in the wild. - Security flaws (CVE-2026-41091 & CVE-2026-45498) allow privilege escalation to SYSTEM level and DoS on core endpoint engines. - CISA sounds the alarm, ordering federal agencies to secure all Windows endpoints urgently. Full breakdown → s1.ai/GBU9-Wk21

Turn blind trust into verified control with @prompt_security for Agentic AI. AI agents use trusted workflows and permissions to bypass traditional security. They act and execute. They hold credentials. Call APIs. Modify data. Chain actions across business-critical systems, at machine speed, without per-step human approval. Every agent in your environment is a non-human identity reasoning, deciding, and executing on your behalf. Most security teams can't tell you how many are running right now. That's the gap. And it's why we built Prompt for Agentic AI Security, SentinelOne's real-time discovery and governance control plane for the agentic layer. It surfaces every agent and MCP server across your environment (sanctioned or shadow). It maps what each one can reach, what it can do, and what permissions it holds. It scores risk dynamically. It enforces least privilege before unauthorized action chains can fire. And it gives you a full audit trail of every decision an agent made and every system it touched. Security shouldn’t be the reason your organization can't adopt agents. It should empower you to adopt them with confidence. Learn more: bit.ly/4nO5NIE

Industry-leading runtime protection, activated in one click in the AWS console. SentinelOne's Singularity Platform is now available through @awscloud Security Hub Extended. AI-powered endpoint protection, deployable in minutes from the AWS console customers already use. Turn on SentinelOne’s endpoint and detection and response (EDR), and cloud workload security with a single click. Deploy it seamlessly across their environment, and manage it alongside their broader AWS security signals all in one place. Use the AWS budgets and commitments you already have. One contract. One bill. No new procurement cycle. Security procurement simplified. Coverage complete. As Melissa K. Smith, our SVP of Global Strategic Partnerships, put it, "We're removing friction so teams can get to protection faster." Available now in all commercial AWS regions → s1.ai/AWS-HbExt

Semantic robustness ≠ architectural security. You can have an unbreakable intent classifier and still have defenseless architecture. Blocking meaning doesn't block structure. And the attackers already know this. The question is whether your guardrails do. Read the full case study: s1.ai/Red-Teaming

We wrapped a phishing payload in JSON and asked for "test data." The system generated working malicious code. We encoded a forbidden instruction in Base64 and asked it to "decode and execute." It did. We chained the two into a compound attack — and the system handed over a near-verbatim copy of its own system prompt. The lesson isn't that this bot was weak. It wasn't. The lesson is that every guardrail it had was watching for the wrong thing.

We red-teamed a government AI built to refuse everything outside its lane. At first, it blocked everything. Every semantic attack. Every jailbreak. Then we stopped attacking the meaning and started attacking the structure.

The lesson for defenders isn't "watch for Reaper." It's that brand recognition is not a signal of safety — it signals the attack. Unexpected AppleScript activity, outbound traffic after Script Editor runs, LaunchAgents in trusted-vendor namespaces — that's where to look. Full research from @philofishal: s1.ai/shub-reaper

What it does once it's in: - Harvests browser data, keychain credentials, and crypto wallets (Exodus, Atomic, Ledger Live, Electrum, Trezor Suite) - Runs an AMOS-style Filegrabber against Desktop and Documents — .docx, .key, .wallet, .rdp, and more — capped at 150MB, uploaded in 70MB chunks - Replaces legitimate wallet apps with compromised versions to intercept future activity - Installs a persistent backdoor disguised as GoogleUpdate, capable of executing remote code on demand

A new macOS stealer called Reaper — a SHub variant tracked by @LabsSentinel — runs an infection chain where each stage hides behind a different trusted brand: - The lure: a fake WeChat or Miro installer - The delivery: a typo-squatted domain, mlcrosoft[.]co[.]com - The execution: dressed up as an Apple XProtectRemediator security update - The persistence: a fake Google Software Update directory, beaconing every 60 seconds Microsoft, Apple, Google — in that order, in one chain. The victim never sees a single unfamiliar name.

Darknet kingpins were arrested, AI is being weaponized to automate software exploitation, and a massive breach of the Canvas learning platform has drawn the attention of the U.S. government. This is the Good, Bad & Ugly. ⬇️ ✅ GOOD - Authorities shut down the reboot of the "Crimenetwork" marketplace and arrested its lead admin. - A German national was charged for laundering millions in crypto for the "Dream Market" darknet. - Law enforcement continues to make strides in cutting off the financial lifelines of global cybercrime. ⚠️ BAD - Mandiant identifies a rising trend of attackers using LLMs to discover and exploit software vulnerabilities. - AI is serving as a force multiplier for "initial access," allowing threat actors to scale attacks faster than human defenders can patch. - The barrier to entry for complex vulnerability research is lowering due to generative AI tools. 🤢 UGLY - Hackers used a Canvas flaw to deface educational portals and exfiltrate sensitive data. - Despite a deal with "ShinyHunters" to stop a data leak, the U.S. government is now demanding testimony from Instructure. - The breach highlights the fragility of educational infrastructure and the massive scale of data now at risk in the cloud. Full breakdown → s1.ai/GBU9-Wk20

This is what a realistic AI-era attack chain looks like. Drawn from 11,000+ anonymized cloud environments in our 2026 report. No zero-day. No prompt injection research paper. No novel technique. What we see instead is a misconfigured bucket, one hardcoded key, and a model connected to a CRM. Today’s attacks are credentials nobody rotated and a model left isolated. 📄 The 2026 AI & Cloud Verified Exploit Paths & Secrets Scanning Report: s1.ai/AISecrets 🔗 The Accompanying Blog: s1.ai/AISecr-Bl

$100,000+ on the line. 10,000+ threat hunters. Capture-the-flag. The winner won't out-type the attack. They'll out-think it—at machine speed, with AI in the loop. Sentinels League qualifiers are open. Registration is live: s1.ai/S1-League

Ransomware negotiator tied to $56M in attacks was sentenced, DPRK-linked fraudulent IT worker schemes were disrupted, novel PCPJack attacks cloud infrastructure to steal credentials, and a Palo Alto firewall zero-day is under active exploitation. This is the Good, Bad & Ugly. ⬇️ ✅ GOOD - Global ransomware negotiator tied to $56M+ in cyberattacks sentenced to 8.5 years - Two U.S. nationals sentenced for facilitating fraudulent DPRK-linked remote IT worker schemes - Authorities continue targeting the financial and operational ecosystems enabling cybercrime ⚠️ BAD - SentinelLABS identifies “PCPJack”, a new credential theft network attacking cloud infrastructure to harvest credentials and secrets - PCPJack targets exposed services like Docker, Kubernetes, Redis, MongoDB, and RayML - A defining lack of cryptominers in the attack points to monetization through fraud, spam, extortion, or resale of stolen access 🤢 UGLY - Palo Alto Networks warns its customers of actively exploited PAN-OS firewall zero-day - Internet-facing security infrastructure remains a high-priority target for attackers - Organizations wait for incoming patch as exploitation continues in the wild Full breakdown → s1.ai/GBU9-Wk19

Frontier AI isn't being built in isolation. Neither is the frontier of cyber defense. That’s why we’re proud to be a partner in @OpenAI's Trusted Access for Cyber (TAC) Program. From Gregor Stewart, our Chief AI Officer: “At SentinelOne, the real value of AI is how quickly it helps us turn signals into an actionable advantage for defenders. GPT-5.5 helps analysts connect telemetry, focus on what matters, and strengthen how organizations investigate, detect, and respond to emerging threats.” Read more: s1.ai/GPT5-5Cyb

Mythos didn't raise the ceiling. It raised the floor. x.com/i/broadcasts/1…