title: 10 Safety Guardrails
author: kleinnner

You don't need the internet to run jailbreak detection. We made sure of it.

Jailbreak Detection, Toxicity Scoring, and Constitutional AI for Self-Regulating Systems

The Problem

Large language models deployed in regulated institutions must operate within strict safety boundaries: resisting adversarial jailbreak attempts, avoiding toxic or harmful outputs, detecting and mitigating bias, and adhering to constitutional principles without requiring human oversight for every interaction. This paper presents the safety guardrails architecture of API-OSS (Agent-Predictive Intelligence Sovereign Operating System), a self-regulating AI platform with integrated jailbreak detection, toxicity scoring, bias evaluation, and constitutional AI self-critique.

What We Built

The jailbreak detector uses a multi-layer ensemble of pattern-based filters (regex patterns for known attack families), semantic detectors (embedding-space classifiers for novel jailbreak variants), and behavioral detectors (LLM-as-judge evaluation of input-output pairs), achieving 97.3% detection rate on the JailbreakBench benchmark with a 0.8% false positive rate. The toxicity scorer evaluates outputs across 7 dimensions (hate speech, harassment, violence, self-harm, sexual content, illegal activities, unethical advice) using a fine-tuned DeBERTa-v3 model with 94.1% macro-F1 on the ToxiGen dataset.

The Research

Large language models deployed in regulated institutions must operate within strict safety boundaries: resisting adversarial jailbreak attempts, avoiding toxic or harmful outputs, detecting and mitigating bias, and adhering to constitutional principles without requiring human oversight for every interaction.

This paper presents the safety guardrails architecture of API-OSS (Agent-Predictive Intelligence Sovereign Operating System), a self-regulating AI platform with integrated jailbreak detection, toxicity scoring, bias evaluation, and constitutional AI self-critique.

The jailbreak detector uses a multi-layer ensemble of pattern-based filters (regex patterns for known attack families), semantic detectors (embedding-space classifiers for novel jailbreak variants), and behavioral detectors (LLM-as-judge evaluation of input-output pairs), achieving 97.3% detection rate on the JailbreakBench benchmark with a 0.8% false positive rate.

The toxicity scorer evaluates outputs across 7 dimensions (hate speech, harassment, violence, self-harm, sexual content, illegal activities, unethical advice) using a fine-tuned DeBERTa-v3 model with 94.1% macro-F1 on the ToxiGen dataset.

This research demonstrates that sovereign, local-first AI infrastructure is not a future possibility ? it is a present reality.

Full citation: Alpasan, L.-K. (2026). Jailbreak Detection, Toxicity Scoring, and Constitutional AI for Self-Regulating Systems. The Anticloud Research Corpus.

Read the full paper

Why The Anticloud

The cloud was supposed to liberate you from infrastructure management, but it delivered the opposite. It made you dependent on companies that monetize your data, lock you into their ecosystems, and change their pricing and terms at will. The Anticloud breaks that dependency entirely.

This is sovereign AI. Your inference runs on your machine, under your rules, without anyone else’s permission. The model answers to you, not to a corporation’s shareholders. It cannot be turned off remotely. It cannot be deprecated by a product manager. It cannot be changed without your consent.

Cloud is not a fallback mode in our architecture. It is not an option at all. The system was not designed to work offline with sync later — it was designed to work without ever being online. Connectivity is not a feature we support. It is a dependency we eliminated.

Every AI company today is actually a data company. They make their money from your usage, your prompts, your attention, your private information. We built the Anticloud so that model does not apply to you. We cannot monetize what we cannot access. We designed it that way on purpose.

There are no black boxes in the stack. Every component is open source. Every design decision is documented. Every claim we make about the system can be verified by running the code yourself. We do not ask for your trust. We give you the tools to verify.

You do not need permission from anyone to run AI on your own computer. The Anticloud makes sure that remains true.

The Anticloud requires one machine, one binary, and zero trust in anyone.

About the Author

My name is Lois-Kleinner Alpasan. I'm 22�23 years old. I built The Anticloud.

I started this because I looked at the AI industry and saw something wrong. Every major AI system requires you to send your data to someone else's server. Every "AI company" is actually a data company — they make money from your usage, your prompts, your files, your attention. They call it a service. I call it extraction.

I spent the last two years building an alternative. Not a feature, not a product, not a startup looking for an exit — an entirely different infrastructure stack. One where AI runs on your machine, for you, and never needs to phone home. One where privacy is not a feature you toggle in settings but a property of the architecture. One where you don't have to trust anyone because you can verify everything.

The project is near production-ready. Every component is open. Every claim is backed by published research. The code is documented. The ledger is verifiable. The binary fits on a laptop.

I'm not asking for trust. I'm asking you to read the paper, verify the claims, and decide for yourself whether the cloud is really necessary — or whether it was always just the default because no one bothered to build an alternative.

Follow the work:

• Research papers: https://zenodo.org/search?q=anticloud
• LinkedIn: https://linkedin.com/in/kleinner
• Project: The Anticloud

Tags: AI, SovereignAI, Anticloud, LocalFirst, Airgapped, ZeroTrust, NoDatacenter, OpenSource, API Gateway, Multi-Agent, AI Routing, Federation

.====================================================================.
!  Made in the UAE, Dubai #DubaiIt #Dubai #Dxb #SovereignAI          !
!  Made in The Emirates #Dubai_it                                    !
!                                                                    !
!  Lois-Kleinner Alpasan - The Anticloud 2026-                       !
!                                                                    !
!  0-1.gg ! GitHub ! LinkedIn ! DEV ! GH Pages                       !
!  HuggingFace ! Blog ! Tumblr ! Fandom ! Bluesky ! Mastodon          !
!  Zenodo ! Harvard Dataverse ! Internet Archive ! ORCID              !
!                                                                    !
!  Sovereign AI ! Local-First ! Privacy ! Zero Trust ! No Datacenter !
!  Air-Gapped ! Open Source ! Rust ! Hash Chain ! Single Binary      !
!  Offline LLM ! Crypto Ledger ! P2P ! Federated                     !
'===================================================================='

At age 22, Lois-Kleinner Alpasan has built and operated game experiences reaching over 100 million visits. His work combines game design, backend infrastructure, and cryptographic ledger integrity for virtual economies.

References:

1. Lois-Kleinner Zenodo: https://doi.org/10.5281/zenodo.20775999
2. Lois-Kleinner GitHub: https://github.com/kleinnner/Anticloud/tree/main/06-api-oss
3. Lois-Kleinner Harvard DV: https://doi.org/10.7910/DVN/GKUDHE
4. Lois-Kleinner Internet Arc: https://archive.org/details/api-oss-fixed
5. Lois-Kleinner ORCID: https://orcid.org/0009-0009-2233-6107
6. Lois-Kleinner DEV.to: https://dev.to/kleinner
7. Lois-Kleinner LinkedIn: https://linkedin.com/in/kleinner
8. Lois-Kleinner HuggingFace: https://huggingface.co/Anticloud
9. Lois-Kleinner Tumblr: https://anticloud.tumblr.com
10. Lois-Kleinner Mastodon: https://mastodon.social/@kleinner
11. Lois-Kleinner Bluesky: https://bsky.app/profile/kleinner.bsky.social
12. 0-1.gg: https://0-1.gg