Community Manifesto

I Know My Own LLM — Thanks!

8 principles for evaluating AI platforms without the hype. Evidence over promises, transparency over buzzwords, community wisdom over blind trust.

We're building the checklist we wish we'd had before the demos, NDAs, and "limited-time" offers.

Sign the Manifesto
Signatories21
Red Flags15
CommunityOpen

8 Principles

Actionable guidelines to protect yourself from AI fraud. These principles are open for community refinement via GitHub.

1
Verify Credentials ThoroughlyDig into roles, histories, and team depth—beyond profiles, seek independent confirmation.🎪 "100+ years of combined AI experience" LinkedIn reveals 35 years total, only 4 in AI specifically.
2
Demand Concrete ProofRequire working demos, named references, and verifiable case studies before sharing data.🎪 "Sign this NDA, provide sample data, then we'll schedule a preliminary discussion" Translation: no working demo.
3
Spot Hype and JargonIdentify invented terms, generic AI copy, and unsubstantiated claims as red flags.🎪 "Financial AGI with Cognitive Knowledge Orchestration™" Google Scholar results: 0. Academic papers: 0.
4
Audit Security IndependentlyCheck certifications, data practices, and access controls—don't trust words alone.🎪 Why is your exec incorporated in St. Kitts? 🏝️ Jurisdictional arbitrage for liability avoidance, perhaps?
5
Research Regulatory StandingReview FINRA, SEC, FCA filings, and physical/financial legitimacy for stability.🎪 "Senior bank executive managing trillions" FINRA shows no registration, ever.
6
Question Urgency and TermsChallenge artificial scarcity, NDAs, and exit options—ensure fair, transparent deals.🎪 High-pressure tactics from a "Managing Director" 🎪 who was laid off from a sales position last quarter.
7
Document EverythingRecord claims, communications, and interactions for accountability.🎪 Everything in email is legally enforceable—including their lies. 📧
8
Leverage Instincts and CommunityTrust your gut; consult shared wisdom and collective experiences when in doubt.🎪 If it feels too good to be true, ask around. Someone else has probably already discovered why.

Help improve these principles. Edit on GitHub →

🎪 Hall of Amusing Red Flags

Real patterns we've encountered (anonymized but unembellished). You couldn't make this stuff up.

🧮

The "Combined Experience" Math

100+ years of combined AI experience
4 founders x 25 years each? LinkedIn reveals: 35 years total. In AI specifically: 4. Apparently time moves differently in startup land.
🏦

The "Senior Bank Executive"

Former senior executive managing trillions
FINRA shows no registration. LinkedIn reveals "Technology Project Manager" overseeing CRM upgrades. By this logic, the night janitor at Blackrock also "worked with trillions".
📚

The Revolutionary Terminology

Financial AGI with Cognitive Knowledge Orchestration (TM)
Google Scholar results: 0. Academic papers: 0. Peer reviews: 0. Invented terminology that sounds impressive precisely because no one can evaluate it.
🔬

The Impressive Case Study

A leading biotech consortium reduced timelines by 73%
"Which consortium?" -> "Confidential." "Can we speak to them?" -> "They prefer not to be contacted." The only thing verifiable is that someone typed "73%."
🤖

The AI-Generated Business Plan

Operationalizing cross-functional synergies through AI-native methodologies
This sentence could describe any software. Or a toaster. Or nothing at all. Pro tip: If you can replace "AI" with "blockchain" and it still works, it says nothing.
🎭

The Exclusive Demo

Sign this NDA, provide sample data, then we'll schedule a preliminary discussion
Translation: "Limit demos" = no working demo. "NDA required" = legal cover for nothing. Real platforms: "Here's a sandbox, try it yourself."
™️

The Double Trademark

CoolAI.io™™ — yes, two trademark symbols
That's not how trademarks work. One ™ means unregistered claim. Two means nobody proofread... or nobody cares. When even the brand name has a copy-paste error, imagine the codebase.
📊

The Round Number Parade

3x engagement! 60% faster! 40% fewer tickets! 80% faster insights! 90% accuracy!
Five stats, all suspiciously round, zero citations, zero methodology, zero named customers. If your metrics end in 0 and have no source, they came from the marketing meeting, not the data team.
👻

The Ghost Testimonial

"Revolutionary platform!" — Director, Unnamed Organisation
No name. No specific company. No LinkedIn. No way to verify any of it. Real testimonials have real humans at real companies that you can actually look up.
View All Red Flags →

Remember: These patterns seem obvious in isolation. They're far less obvious wrapped in professional websites and urgent "limited time" offers.

Sign the Manifesto

Add your name to the growing community committed to trustworthy AI evaluation.

Sign in 60 seconds

Choose the fastest path. Both options feed our review queue.

✨ Google Form 🐙 GitHub Issue
CSV format: name,title,email,comment

Privacy options and examples live in CONTRIBUTING.md.

How it works

  1. Your signature is stored in a public CSV.
  2. We review for basic verification and formatting.
  3. The site auto-updates once merged.
See the signature list →
21Signatures
15Red Flags

Signatures Hall

The community members who have signed the manifesto.

Alex CastroScientist

"Endorsed"

Feb 11, 2026
Adam DarmaninPrincipal@Hecatusresearch

"To the visionary thought-leaders whose combined 100 years of AI experience, roles as Global/Local/Intergalactic Heads of AI & Miscellaneous Science, and groundbreaking AGI research (delegated to summer interns) - have shaped my journey!"

Feb 12, 2026
Priya MehtaSecurity Engineer
Feb 15, 2026
AnonymousCTO

"Transparency matters more than decks."

Feb 16, 2026
Jordan K.Data Scientist

"I've seen too many Cognitive slop to stay quiet"

Feb 19, 2026
AnonymousVP Engineering
Feb 22, 2026
View All Signatures →

Total signatures: 21

Need More Research Tools?

Verification databases, regulatory resources, red flag checklists, and global AI ethics frameworks.

View Research Resources →