FRAI

Helping startups integrate AI the right way: safe, compliant & fast.

[FRAI @openFRAI]

Building https://t.co/TvrmrxBca5, a new framework to help startups build responsible AI fast and right.

Just getting started:
• Free CLI live
• Open-source
• Landing up

Now building the SaaS dashboard with real data, compliance scoring, and auto docs.

Follow for updates.

[FRAI @openFRAI]

We built FRAI for that half. It opens your site's chat the way a visitor would and reads whether it says it's AI.

Free check: https://t.co/Ag3qhtsC6c
Anthropic's post: https://t.co/W3aiFDKVOa

[FRAI @openFRAI]

Anthropic now watermarks Claude's text for the EU AI Act.

Invisible to people. Readable by a machine with the key.

That covers the model. It doesn't cover your support bot.

The customer typing into it still has to be told it's AI. That part is on whoever ships the chat.

[FRAI @openFRAI]

Dario Amodei's new essay: AI safety claims should be checked, not trusted. Anthropic is letting outside evaluators in.

Right idea. It stops at a handful of labs.

The chatbots your customers talk to are still checked by whoever ships them.

We built FRAI for that gap

[Dario Amodei @DarioAmodei]

We Must Pace the Frontier: I’ve written a new essay on why the AI industry should slow down, with a three-part plan for doing so.

Anthropic is unilaterally committing to the first of these steps. We’ll provide third-party evaluators with permanent, employee-level access to our systems, so that they can verify adherence to our safety measures, report on incidents, and assess models’ alignment during training.

You can read the full post here: https://t.co/OGyPb7yaYt

[FRAI @openFRAI]

@sebuzdugan 🔥👀

[Sebastian Buzdugan @sebuzdugan]

Most AI features ship with zero risk review

So I built a gate that refuses to let them:

npx frai-gate blocks your spec until 7 responsible AI checks are answered: risk tier, data, oversight, evals, bias, monitoring, transparency

Empty answers = CI fails

[FRAI @openFRAI]

All of it open:

https://t.co/d4RPEHeqTy
https://t.co/MNmiDnPzrL

[FRAI @openFRAI]

Nobody wants to do AI compliance.

So we spent a year making it something you run, not write.

Microsoft gave us $5,000 in credits last August.
This month they added $20,000 more.

Year one: 5 open tools, all free.
Next month: new partners, simpler FRAI, real impact

[Sebastian Buzdugan @sebuzdugan]

Four open-source AI models just did a real job for me, total cost $1.79. One even wrote code to double-check its own work. I gave all four the same task on Hyperagent and fact-checked every word. The results genuinely surprised me. #HyperagentPartner

[FRAI @openFRAI]

@sebuzdugan @getmaximai awesome tool🤝

[Sebastian Buzdugan @sebuzdugan]

Want to land freelance clients with AI and cross $10k monthly?

I spent $1,000+ on courses and PDFs
Most of it was useless

I kept the part that pulls clients and closes deals

Turned it into a 38-page guide

I’m dropping it FREE
But not for everyone

1.Follow @sebuzdugan
2.Reply “MAGNET”

I’ll DM it to you

[FRAI @openFRAI]

@sebuzdugan MAGNET

[FRAI @openFRAI]

@sebuzdugan 👀

[🍓🍓🍓 @iruletheworldmo]

codex 5.3 is terrifying.

[FRAI @openFRAI]

https://t.co/N32Hvs0e7y

[FRAI @openFRAI]

First results from the FRAI Benchmark!

🚀 Tested 200+ safety vectors on SOTA models (behind Azure Enterprise Filters).

🏆 gpt-5.2 (95%) 🥈 Kimi-K2 (87%)

Fully Open Source & Reproducible. Repo 👇

Do you think we should add more models to see how Responsible they really are?

[FRAI @openFRAI]

are you ready for what we are about to ship around clawdbot 👀

agent are no longer toys
they touch files browsers real accounts

chat becomes an execution surface

at @getfrai we are building the safety layer so this stays powerful not dangerous 🔐

[FRAI @openFRAI]

Stealth mode is dead

We're building the world's first open-source AI compliance benchmark

→ Every test case: public
→ Every model score: public
→ Every methodology: public

GPT vs Claude vs Gemini on bias, safety, jailbreaks

Launch in 3 weeks

[FRAI @openFRAI]

If you're building with AI and ignoring compliance, read this.

It’s not just about regulations, it’s about speed, trust, and traction.

5 reasons you should care about responsible AI 👇🧵