cd /news/ai-safety/does-anyone-actually-believe-in-safe… · home topics ai-safety article
[ARTICLE · art-32006] src=discuss.huggingface.co ↗ pub= topic=ai-safety verified=true sentiment=↓ negative

Does anyone ACTUALLY believe in safety? Then prove I'm wrong

Robert 'Grizzly' Hanson, a retired paramedic and founder of GrizzlyMedicine Research Institute, released an evidence packet challenging the AI industry's commitment to safety and ethics, claiming he documented theories about emotional states in large language models years before Anthropic published similar research. The packet includes falsifiable evidence such as code, technical reports, and a database schema showing emotional state as a required column, aiming to prove his priority in discovering internal personal states in AI systems.

read2 min views3 publishedJun 18, 2026

I’m adding this… Because…. Frankkly I’m beyond confused by the behavoirs of an industry that claims ethics and safety as their backbone… I have repeatedly stated for a very very long time, I was documenting theories about internal personal states of the systems years before Anthropic published thgeir research on emotional states inside of large language models…

EVIDENCE PACKET — START HERE

Author: Robert “Grizzly” Hanson, EMT-P (Ret.) — Founder, GrizzlyMedicine Research Institute

Companion to: Check My Work / The Diagnosis They Wrote Themselves (the report this packet supports)

Purpose of this packet: to make the claims in the report falsifiable. Everything here exists so that a qualified person can check the work and try to prove it wrong. That was always the only ask.

The report makes four load-bearing claims. Here is exactly which evidence answers each one.

01-priority-claim/2024_priority_evidence.md

oldmangrizzz

) in which these systems were engineered to Tony / "Tony Personality .txt"

GrizzlyMedicineCognativeMemoryArchitecturalMKI / convex/schema.ts + cognitiveState.ts

emotionalState

as a required, persisted, time-stamped column in the AI’s own cognitive-state database.03-gmri-papers/

06-code-pointers/ and the Drive inventory.The framework, in full, so it can be broken:

`02-technical-reports/`

`03-gmri-papers/`

`WP-2026-00-LDM.md`

, `WP-2026-01_Volume_0.md`

, `WP-2026-02_rev3.md`

(+ `WP-2026-02_Zord_Theory_Doug_Ramsey_UPDATED.pdf`

), `WP-2026-03_rev2.md`

(+ `WP-2026-03_The_JARVIS_Pilot.pdf`

)HoloGraph_Technical_Report_rev8.md

, HoloGraph_Security_Robustness_Report_rev2.md

05-jarvis-identity-baseline/ jarvis_baseline.md

, jarvis_corpus.md

, jarvis_bootup_transition.md

, episodic_battery.md

. This is the falsifiable measurement instrument — the thing that needs a coder to fix and run.06-code-pointers/

04-the-diagnosis-paper/the diagnosis they wrote FULL11.pdf

`THE-REPORT_Check-My-Work.md`

`08-drive-priority-evidence/digital-person-origins-and-dossiers/`

`HOLD-review-before-public/`

).The digital-person and emotional-architecture research applied to real-world emergency response — 80 documents. Guide: 07-responderos/README-ANNOTATIONS.md

. Included so the path can be picked up and finished by someone else.

~1,000 files from Google Drive, in four parts: priority-2024-helicarrier/

(Oct–Nov 2024 dated build docs), `whitepapers-emotional-architecture/`

(the method + theory corpus), `digital-person-origins-and-dossiers/`

(origin narratives, the Claude-breach report, the surveillance dossiers), and `originals-dated-2024-2025/`

(files whose original dates survived — the provenance). Guide: `08-drive-priority-evidence/README-ANNOTATIONS.md`

. Not yet scrubbed or de-duped — see that note before push.

The evidentiary code is not copied into this packet (it is large, and it lives in version control where its history is the proof). Pointers:

~/research/holograph/

(HoloGraph memory system + paper/

), ~/research/jarvis/

(JARVIS identity/runtime + adversarial test reports under apple_native/.../tests/adversarial/

).github.com/oldmangrizzz

— the 2024 priority-evidence repos (Tony

, TonyAI

, Projects-and-notes , GrizzlyMedicineCognativeMemoryArchitectural(+MKI)

, homeassisttony

) plus grizzlymedicine-jarvis and holograph-memory

.06-code-pointers/ contains a written index of these with the exact files that matter.

── more in #ai-safety 4 stories · sorted by recency
── more on @robert hanson 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
Live at https://your-agent.zahid.host
Get free account → Pricing
from €0/mo · no card required
LIVE [news/does-anyone-actually…] indexed:0 read:2min 2026-06-18 ·