We're building the defense factory for all software. Try tCell, our frontier cyberagent, or build your own with the Antigen SDK.
Joined September 2025
- Tweets53
- Following5
- Followers1.1K
- Likes88
Pinned Tweet
Meet Antigen Assets.
Assets gives your security agent superpowers. Antigen can now find stale deployments that access your core systems and triage the effects of new CVEs the moment they drop.
- It understands how every service connects
- Builds a live map of your entire system
- Connects AWS, Okta, GitHub, Vercel, and many others
Every time your infrastructure changes, it runs a targeted pentest, like an expert red-team who’s spent hours studying your architecture.
Antigen retweeted
You can not bring humans to an agent-scale war.
This week was the beginning of a new era in computer security, and in the coming days, we'll see this unfold at scale.
At @antigenco, we're deploying swarms of adversarial agents within America's most sensitive targets.
Our customers require strong data privacy, on-prem deployments, and absolute aggression that matches the most serious and determined adversaries.
Excited we get to play this key role.
This is the most serious 48 hours in cybersecurity history.
php2Shell has left 70% of the internet vulnerable as @NSAGov is advising some American companies to take production systems offline.
@eMudhra , an Indian Root Certificate Authority, relied on weak cryptographic keys (one was literally the number 4) that allows attackers to man-in-the-middle any website.
@LiteLLM had its release pipeline hijacked and became a vector for distributing malware. Attackers stole inference API keys and installed RCE backdoors in v1.107.0.
@Railway exposed customer environment variables like cloud credentials and inference keys. This is part of a model-distillation attack to steal inference output from labs.
@gitlab disclosed a 10.0 / 10.0 CVSS vulnerability that allows arbitrary code execution on every single version since 2014.
@jpmorgan, @BankofAmerica , @WellsFargo , and @MorganStanley have detected unauthorized AI agents operating inside customer accounts. The financial institutions are unable to distinguish legitimate customer activity from AI agents at-scale.
@opencode was the target of a 48-hour agent swarm where its infrastructure was mined for customer API keys.
If you still think bringing human security teams to fight the AI war is a sensible defense strategy, consider yourself pwned.
Antigen retweeted
If you weren't working with @antigenco day before yesterday.... You probably should be by now as part of your security stack.
Good people means a lot when it comes to security.
This is the most serious 48 hours in cybersecurity history.
php2Shell has left 70% of the internet vulnerable as @NSAGov is advising some American companies to take production systems offline.
@eMudhra , an Indian Root Certificate Authority, relied on weak cryptographic keys (one was literally the number 4) that allows attackers to man-in-the-middle any website.
@LiteLLM had its release pipeline hijacked and became a vector for distributing malware. Attackers stole inference API keys and installed RCE backdoors in v1.107.0.
@Railway exposed customer environment variables like cloud credentials and inference keys. This is part of a model-distillation attack to steal inference output from labs.
@gitlab disclosed a 10.0 / 10.0 CVSS vulnerability that allows arbitrary code execution on every single version since 2014.
@jpmorgan, @BankofAmerica , @WellsFargo , and @MorganStanley have detected unauthorized AI agents operating inside customer accounts. The financial institutions are unable to distinguish legitimate customer activity from AI agents at-scale.
@opencode was the target of a 48-hour agent swarm where its infrastructure was mined for customer API keys.
If you still think bringing human security teams to fight the AI war is a sensible defense strategy, consider yourself pwned.
Black October is upon us. Today we are calling for every security team to roll-out their Defense Factory: dedicated security agents that continually probe your systems for vulnerabilities, autonomously remediate, and re-test to verify the patch is successful.
In the era of agentic attackers, it is no longer sufficient to bring humans to an AI war.
This is the most serious 48 hours in cybersecurity history.
php2Shell has left 70% of the internet vulnerable as @NSAGov is advising some American companies to take production systems offline.
@eMudhra , an Indian Root Certificate Authority, relied on weak cryptographic keys (one was literally the number 4) that allows attackers to man-in-the-middle any website.
@LiteLLM had its release pipeline hijacked and became a vector for distributing malware. Attackers stole inference API keys and installed RCE backdoors in v1.107.0.
@Railway exposed customer environment variables like cloud credentials and inference keys. This is part of a model-distillation attack to steal inference output from labs.
@gitlab disclosed a 10.0 / 10.0 CVSS vulnerability that allows arbitrary code execution on every single version since 2014.
@jpmorgan, @BankofAmerica , @WellsFargo , and @MorganStanley have detected unauthorized AI agents operating inside customer accounts. The financial institutions are unable to distinguish legitimate customer activity from AI agents at-scale.
@opencode was the target of a 48-hour agent swarm where its infrastructure was mined for customer API keys.
If you still think bringing human security teams to fight the AI war is a sensible defense strategy, consider yourself pwned.
For teams that don't know what this means or how to implement it. We're here to help.
Talk to us today: cal.com/team/antigen/demo?ov…
Antigen retweeted
Meet Antigen Assets.
Assets gives your security agent superpowers. Antigen can now find stale deployments that access your core systems and triage the effects of new CVEs the moment they drop.
- It understands how every service connects
- Builds a live map of your entire system
- Connects AWS, Okta, GitHub, Vercel, and many others
Every time your infrastructure changes, it runs a targeted pentest, like an expert red-team who’s spent hours studying your architecture.
Stay continually secure and launch your first pentest at antigen.sh/
OpenAI turned its most capable cyber models against their own infrastructure to continuously attack their systems.
In the age of AI-powered attacks, defense factories that find vulnerabilities, remediate them, and verify the patch autonomously are the only way to stay secure.
We've spent months building this with major healthcare systems and cloud providers. Today, we’re making it available to everyone.
> Find: Continuously probes your systems to identify vulnerabilities
> Fix: Writes the fixes and opens PRs for review
> Verify: Retests every patch until the vulnerability is gone
Find. Fix. Verify. Repeat.
Antigen only considers a vulnerability patched once it is proven.
It integrates with your deployment pipeline, and once the fix rolls out to production, reattempts the original exploit.
If some path to your critical data still exists, the vulnerability reopens with new evidence, and the loop starts again.
We call this loop the defense factory.
Build your defense factory with Antigen.
Find a time to chat with our team: cal.com/team/antigen/demo
Antigen retweeted
We’re excited to share Antigen publicly. The past few months, we’ve worked with healthcare systems and major cloud providers to build the modern defense factory.
Our systems have protected Social Security numbers of thousands of Americans, secured workloads carrying sensitive information, and protected state and local government data.
We've reengineered security infrastructure from core primitives: Antigen deploys in your system, attacks continuously, ships fixes, and loops until you're truly secure.
Book a demo to see what we find!
Antigen retweeted
Excited to be working with health insurance networks, payment providers, and critical power/water utilities to keep the systems around us secure.
My favorite part? Antigen drafts the suggested fix for every vulnerability it detects, allowing your security engineers to focus on hard design problems and not silly config files.
Use the Antigen SDK to run penetration tests as part your CI, and stay continually secure with every deployment.
docs.antigen.sh/docs/sdk
See what tCell finds in your environment. We’re offering a free evaluation pentest of your approved targets.
Book yours: cal.com/team/antigen/demo
