Anthropic Warns: Advanced Cyber Capabilities Spreading to More AI Models
Anthropic has published an analysis of the GLM-5.3 AI model, indicating that advanced cyber capabilities are now spreading to an increasing number of AI systems.
What happened?
Anthropic has published an analysis of the GLM-5.3 AI model, developed by the Chinese company Zhipu AI. The analysis shows that advanced cyber capabilities, such as the automated detection and exploitation of software vulnerabilities, are now being integrated into a wider range of AI models. Anthropic previously reported that its own Claude Mythos Preview model was capable of building advanced cyber-exploits, and the company forecasted that the technology would spread rapidly to other actors.
Key facts
| Publiceringsdatum | 29 september 2026 |
|---|---|
| Granskad AI-modell | GLM-5.3 (Zhipu AI) |
| Identifierade sårbarheter | Över 10 000 i Project Glasswing |
”The rapid rate of improvement in AI suggested to us that this ability would eventually proliferate to many other models, making it much easier for malicious cyber actors to launch highly impactful cyberattacks.”
Why it matters
The proliferation of advanced cyber capabilities across more AI models is altering the digital threat landscape. As tools for automated cyberattacks become available via more sources, the pressure on IT defenders to rapidly identify and remediate vulnerabilities in critical software continues to rise.
Who is affected?
AI model developers, IT security experts, cybersecurity firms, and security officers at government agencies and corporations are affected. Furthermore, providers of critical infrastructure are impacted, as they must secure their systems against automated cyberattacks.
Impact on the EU
Security issues concerning advanced AI and cyber capabilities fall directly under the EU AI Act, which imposes strict risk management requirements on models that pose systemic risks. EU-based organizations are affected both by the identified vulnerabilities and by future oversight requirements.
What else you should know
Anthropic originally distributed selected functions of Claude Mythos via Project Glasswing to authorized security experts. This approach aimed to detect and fix flaws before corresponding AI models were widely distributed on the open market.
Quick answers about this story
Vad har hänt?
När hände det?
Varför spelar det roll?
Påverkar det EU?
The link opens in a new window and leads to the publisher's own site.
Källan har spårats automatiskt från utgivaren via Aheadlines signalkedja.
AI-verktyg i artikeln
Topics
Get similar news straight to your inbox
The reader's room
Send in a question or an addition. The newsroom reads everything before it's published and replies when relevant. No AI-generated text – just people.
Sign in to submit a comment or question.
Read the article through your role
- Decide whether this affects strategy over 6–12 months or is just noise.
- Discuss with leadership: do we own the right question or does ownership need to move?
- Ask: what risk are we taking by NOT acting on this this quarter?
Generated angle — not editorial analysis of "Anthropic Warns: Advanced Cyber Capabilities Spreading to Mo"