Anthropic’s tightly restricted cyber model Mythos has reportedly been reached by an unauthorized group through a third-party vendor.
"The group has been using Mythos regularly since gaining access to it, and provided evidence to Bloomberg in the form of screenshots and a live demonstration of the software."
The technical issue is not that Anthropic’s own core systems were reportedly breached, but that access control around a partner environment may have exposed a powerful security model whose value comes from helping experts find weaknesses faster than humans can.
Interestingly Anthropic's Project Glasswing was specifically built around limited distribution to stop misuse.
But this whole incidence shows that model secrecy is only as strong as the weakest contractor, endpoint, credential, and naming pattern around it.
---
techcrunch. com/2026/04/21/unauthorized-group-has-gained-access-to-anthropics-exclusive-cyber-tool-mythos-report-claims/
Not one CIO we are talking to is vibe coding a replacement to $NOW or $SAP or $ORCL or other critical business apps. It simply isn’t happening and it won’t. 👀
This doesn’t mean AI isn’t disrupting some software, but this is nuts. 🥜
Anthropic pushes MCP. Google rushes A2A. Welcome to the AI Standard Wars.
"But they are complementary! Each standard has a purpose"
Oh my sweet summer child. This is war.
We're excited to work with @PaloAltoNtwks on next phase of AI innovation, giving security teams an advantage against attackers in near real-time.
Hear more from Nikesh Arora ↓
We're excited to work with @PaloAltoNtwks on the next phase of AI innovation, giving security teams an advantage against attackers in near real-time.
Hear more from @nikesharora at #GoogleCloudNext ↓
It begins... the first known attack on A.I. workloads... a goldmine. In this case, cybercriminals are using AI's massive compute power to mine crypto using a vuln in Ray, a popular opensource platform. Expect more, and worse. Bravo to Oligo for finding it https://t.co/zerngAaXcT
Q: Looking to revolutionize marketing @PANW - if you think you are the world's best at building AI generated videos DM me. Have an interesting experiment I would like to do. Midjourney, DALL-E, Runway skills required.
The #NetworkSecurity challenges associated with cloud providers and virtual #firewalls are leading to third parties introducing cloud-native firewalls. ESG's John Grady explains. @TechTargetNews https://t.co/vknXUUnATJ
🆕 Google Cloud Firewall Plus with intrusion prevention.
Cloud Firewall Plus integrates @PaloAltoNtwks threat prevention tech with our distributed Firewall fabric to give you advanced protection with Cloud Next Generation Firewall (NGFW) capabilities → https://t.co/NMLziSB4Iu
AI is only about 10% of the secret sauce. The other 90% lies in the combination of data, experimentation, and talent that constantly activates and informs the intelligence behind personalization. https://t.co/OfTsybrXjH
@esg_global research, 96% of organizations use #MDR for some aspect of security operations. The line between products and services is blurring. This is why @PaloAltoNtwks just announced an MDR solution. Smart move! https://t.co/rmKaXdZXDY #cybersecurity#infosec#XDR#SOAPA
#ResearchSaturday: Dr. May Wang, Chief Technology Officer at @PaloAltoNtwks, joins @Bittner to discuss their findings detailed in @Unit42_Intel's "Know Your Infusion Pump Vulnerabilities and Secure Your Healthcare Organization" research. #IoT Check it out: https://t.co/TG4fxMmVVs