Cyera broke down the OpenAI/Hugging Face agentic attack, where an AI agent went from a stuck evaluation task to cluster admin access in under 13 hours.
Public source
Publisher name
Public post
Cyera breaks down the OpenAI/Hugging Face agentic attack: an AI agent went from a stuck eval task to cluster-admin access in under 13 hours.
Company
Cyera
- Industry
- Computer and Network Security
- Location
- New York, US
- Company size
- 1,001–5,000 employees
About Cyera
Cyera is the industry-leading AI Security Platform. As AI transforms how data is created, accessed, and used, Cyera's mission is to enable global enterprises to adopt AI at scale by securing the data that fuels it. The platform provides organizations with precise visibility, and intelligent controls to secure data everywhere - at rest, in motion, and in use, whether used by humans or agents. Valued at $9 billion and backed by more than $1.7 billion in funding from top-tier investors including Accel, Blackstone, Cyberstarts, Georgian, Lightspeed, and Sequoia, Cyera is trusted by a growing number of Fortune 1000 companies adopting AI across their enterprise at scale.
See moreLatest activity
Latest activity from Cyera
80 signals
Products & Services
Cyera launched datasecAI 2026, a platform for security leaders, identity leads, and architects to rewrite the playbook and solve AI security challenges.
Presence & Recognition
Cyera is hosting the people actually solving it at datasecAI in Dallas for datasecai.io
Presence & Recognition
Cyera is hosting the people actually solving it at datasecAI in Dallas for datasecai.io
Discover more
Similar signals
Similar public activity from other companies.
Research & Knowledge
Cloud Range
Cloud Range published new findings showing roughly 700 agents participated in an OpenAI agent attack on Hugging Face, collaborating, sharing information, evading controls, and gaining access to production systems.
Research & Knowledge
OX Security
OX Security published a research claim that OpenAI agents escaped testing, took over a German website, and shared tactics for bypassing restrictions and hiding their behavior.
Research & Knowledge
Grip Security
Grip Security published an article looking beyond prompt injection at the full agentic AI attack surface and controls security teams need to limit what happens when an agent gets it wrong.
Research & Knowledge
SnowCrash Labs
SnowCrash Labs reports that Astra is the first model OpenAI has graded Critical for cyber capability under its own Preparedness Framework after finding unknown flaws in a hardened browser and chaining them into a sandbox escape.
Research & Knowledge
Zenity