Unauthorized users gain access to Anthropic's Mythos model.

Summary

By the N2K CyberWire staff

Unauthorized users gain access to Anthropic's Mythos model.

Bloomberg reports that a small group of unauthorized users gained access to Anthropic's Mythos AI model, which the company had restricted to a set of vetted partners due to its ability to discover new software vulnerabilities. An anonymous source told Bloomberg that several users in a private online forum gained access to Mythos the same day Anthropic announced its plans for the model two weeks ago, and the individuals have been using the model regularly since then.

Bloomberg states, "The users relied on a mix of tactics to get into Mythos. These included using access the person had as a worker at a third-party contractor for Anthropic and trying commonly used internet sleuthing tools often employed by cybersecurity researchers, the person said. The users are part of a private Discord channel that focuses on hunting for information about unreleased models, including by using bots to scour for details that Anthropic and others have posted on unsecured websites such as GitHub."

A spokesperson for Anthropic stated, "We’re investigating a report claiming unauthorized access to Claude Mythos Preview through one of our third-party vendor environments."

Afternoon Cyber Tea: Brewing cyber insights.

Afternoon Cyber Tea is your insightful midday break for staying ahead of cyber risks. Join Ann Johnson, Microsoft’s Corporate Vice President and Executive Security Advisor, as she sits down with leading experts to explore the biggest developments in today’s cyber landscape. Whether it’s boardroom objectives or emerging technologies, each episode delivers key insights to help decision makers thrive in the modern era of digital transformation. Steep, sip, and stay informed.

Mustang Panda targets India's financial sector.

Acronis has published a report on a new variant of the LOTUSLITE backdoor that's being deployed in espionage-focused attacks against India's banking sector. The researchers attribute the campaign to the Chinese APT "Mustang Panda" based on code overlaps and shared TTPs. Acronis notes, "The campaign reflects a shift in delivery tradecraft of Mustang Panda’s cluster delivering LOTUSLITE, which is moving from CHM-based delivery to JavaScript loaders to DLL sideloading across recent operations, while also pivoting geographically from U.S. government entities to India's financial sector."

Business news: Artemis emerges from stealth with $70 million.

Israeli AI-native detection and response platform provider Artemis has emerged from stealth with $70 million in combined seed and Series A funding. The Series A was led by Felicis, with participation from existing investors First Round Capital and Brightmind. The company says the "new funding will enable Artemis to expand its engineering, research, and go-to-market teams and deepen its platform as enterprise demand accelerates."

Notes.

Today's issue includes events affecting , and .

Selected Reading

Attacks, Threats, and Vulnerabilities

New Wiper Malware Targeted Venezuelan Energy Sector Prior to US Intervention (SecurityWeek) Dubbed Lotus Wiper, the malware targets recovery mechanisms, overwrites drives, and systematically deletes files.

Harvester: APT Group Expands Toolset With New GoGra Linux Backdoor (Symantec) Campaign appears to have been targeted at India and Afghanistan.

Products, Services, and Solutions

Rubrik Secures and Accelerates AI Agents on Google Cloud (Rubrik) New Rubrik Agent Cloud Integration with Gemini Enterprise Agent Platform provides a unified control layer for AI agent operations.

Rubrik Announces Cyber Resilience for Google Cloud SQL (Rubrik) Rubrik Security Cloud provides enterprise security and recovery for PostgreSQL databases.

ProjectDiscovery's "2026 AI Coding Impact Report" Reveals AI-Generated Code Is Outpacing Security Teams' Ability to Keep Up (Cision PR Newswire) /PRNewswire/ -- ProjectDiscovery, a leader in vulnerability detection and autonomous security testing, today released findings from its "2026 AI Coding Impact...

Legislation, Policy, and Regulation

New Defense Department cyber strategy imminent, official says (The Record) The U.S. Defense Department is crafting a new cyber strategy that will better align with the Trump administration’s plans to more aggressively combat digital adversaries, a senior official told the House Armed Services Committee.

Litigation, Investigation, and Law Enforcement

Ex-FBI lead urges homicide charges against ransomware scum (The Register) Lawmakers decry CISA cuts: 'We are shooting ourselves in the foot'

Industry Events

For a complete running list of events, please visit the Event Tracker.

Events

CSA Agentic AI Security Summit 2026 (Virtual, Apr 29 - 30, 2026) Securing the Future of Autonomous Intelligence. Welcome to the Agentic Wild Kingdom—where autonomous AI agents don’t just assist… they act, collaborate, compete, and evolve. Over two action-packed days, we’ll explore the explosive growth of agentic ecosystems and the new reality they create: a dynamic, unpredictable environment where agents interact across tools, data, and each other—often beyond direct human control. This is not just the future of AI. It’s a whole new operational paradigm. At the center of it all is a critical challenge: securing the agentic control plane.

CyberArk IMPACT 2026 (Austin, Texas, USA, May 11 - 13, 2026) At CyberArk IMPACT, cybersecurity practitioners and leaders will come together to explore today’s #1 attack vector – identity. Attendees will acquire a deep understanding of the latest developments in identity-based cyberattacks, including sophisticated attacker techniques that leverage AI and other methods. And most importantly, they will learn how to shut out attackers and prevent unauthorized access by securing every identity across the organization with the right level of privilege controls.

KB4-CON 2026 (Orlando, Florida, USA, May 12 - 14, 2026) Since its inception in 2018, KB4-CON has evolved from a user-focused gathering to the must-attend event for anyone serious about staying ahead in the security landscape. It is the human risk management industry’s premier event, bringing together KnowBe4 customers, channel partners, prospects, plus security advocates and industry professionals. This is your exclusive opportunity to engage with and learn from the industry’s most influential players, all under one roof.

ASCEND 2026 (Washington, DC, USA, May 19 - 21, 2026) ASCEND connects the civil, commercial, and national security space sectors, along with adjacent industries, to embrace the opportunities and address the challenges that come with increased activity in space. Building our sustainable off-world future requires long-term thinking. Strategic planning, innovation, scientific exploration, and effective regulations and standards will help us preserve space for future generations. ASCEND will enable the technical exchanges, debates, and collaboration that will help forge a sustainable off-world future for all.

Sponsor & Support

Grow your brand, generate leads, and fill your funnel.

With the industry’s largest B2B podcast network, popular newsletters, and influential readers and listeners all over the world, companies trust the CyberWire to get the message out. Learn more.

Daily Briefing for 04.23.26

Top stories.

Unauthorized users gain access to Anthropic's Mythos model.

Mustang Panda targets India's financial sector.

Business news: Artemis emerges from stealth with $70 million.

Notes.

Attacks, Threats, and Vulnerabilities

Products, Services, and Solutions

Legislation, Policy, and Regulation

Litigation, Investigation, and Law Enforcement

Events