Friday, 25 September 2026European Markets
ExploreSearch
ViaNews EU
Source trace. Via News points to the documents behind its reporting and shows what we drew from each — so you can check any claim. How we source
Source document

A new Anthropic model found security problems ‘in every major operating system and web browser’

View original at theverge.com
The Verge AI - Consumer Ai Title: A new Anthropic model found security problems ‘in every major operating system and web browser’ Date: 2026-04-07 18:00 Source: https://www.theverge.com/ai-artificial-intelligence/908114/anthropic-project-glasswing-cybersecurity <figure> <img alt="Photo collage showing mouse cursors cir…
Opening lines of the source · short snapshot — read the full document at the original

What we drew from this source

The claims Via News extracted from this document. We point to the source; we don't replace it.

  • Anthropic is not currently planning to publicly release Claude Mythos Preview due to security concerns

    60% confidence

  • Project Glasswing is a way for large companies, and potentially even the government, to flag vulnerabilities in their systems with virtually no human intervention

    60% confidence

  • Claude Mythos Preview found security problems in every major operating system and web browser

    60% confidence

  • Claude Mythos Preview is not currently planned for public release due to security concerns

    60% confidence

  • The model will ideally give cyber professionals advanced vulnerability detection capabilities

    60% confidence

  • The new AI model found security problems in every major operating system and web browser

    60% confidence
What we know · the intelligence behind this page
Live from the substrate
What we're seeing
Enterprise AI Agents Scale Up Through Partnerships and Funding, But Data Readiness Lags Ambition
A wave of vertical AI-agent startups (Swarm, Veridox, Avallon AI, DA2, F2, Earthian, Meanwhile, Covecta, Penguin AI, Maisa AI) is being funded and profiled just as major infrastructure players — Microsoft/Mistral, Siemens/NVIDIA, and Manulife/Microsoft — cement enterprise AI governance and compute partnerships. Yet a Google Cloud report shows AI agents still lack access to the majority of company data (only 45% on average), and insider selling at incumbent C3.ai signals investor caution even as adoption intent (100% planned agentic AI use within two years) races ahead of actual data infrastructure.
Our read on the data ›
Signals we're tracking
EPKINLY Regulatory-Clinical Success Cascade
High probability of expanded label indications, additional combination approvals, and competitive positioning strength in follicular lymphoma market. Predicts positive commercial uptake and potential accelerated review for related indications.
Patterns we're watching ›
Where sources disagree
Berkshire Hathaway
Both facts report Berkshire Hathaway's cash position on 2026-01-01 with identical observation timestamps, but claim vastly different values: 380 billion USD vs 400 USD. These cannot both be true for the same entity at the same point in time. The magnitude of the discrepancy (a factor of ~10^9) rules out rounding, unit conversion, or methodological differences.
We flag conflicts openly ›
Recently verified
✓ Checked against the original source
4,984
facts traced to their source — and we flag the ones that don't hold up.
101 entities tracked4,984 facts checked against source5,306 source documents archived
Query this data → isubstrate.com
A new Anthropic model found security problems &#8216;in every major operating system and web browser&#8217; — Source | Via News | ViaNews EU